**** BEGIN LOGGING AT Wed Jun 29 23:59:56 2005
Jun 30 00:33:46 <jacques> yay!!!!!
Jun 30 00:34:00 <jacques> looking for place to plugin soldering iron right now
Jun 30 00:35:46 <NAiL> haha
Jun 30 00:36:00 <NAiL> Gonna remove R64 or R83?
Jun 30 00:36:03 * jacques looks for a tiny fan yo mount in the slug.
Jun 30 00:36:25 <NAiL> I think just a heatspreader will do fine
Jun 30 00:36:29 <jacques> NAiL, I haven't read the full backlogs yet so I dunno the difference between R64 and R83.
Jun 30 00:36:44 <NAiL> R83 seems to be the *proper* way to do it
Jun 30 00:36:58 <rwhitby-away> one is undocumented and works, other is documented but untried
Jun 30 00:37:01 <jacques> I also need to look at the board - is one significantly easier to get to/remove ?
Jun 30 00:37:26 <NAiL> R64 is slightly easier to remove. R83 is crammed between R82 and R81
Jun 30 00:37:28 <jacques> I recall from modding one of my scanners that these thigns are much easier to remove than replace :-)
Jun 30 00:37:47 <NAiL> R64 is actually *dead easy* to remove.
Jun 30 00:37:52 <rwhitby-away> yeah we're only considering removal options
Jun 30 00:38:14 <jbowler-zzz> I had thought that the one to remove was the one just below the R81 pads.
Jun 30 00:38:32 <NAiL> below the R81 pads?
Jun 30 00:38:55 <jacques> I hope all the ppl doing this mod in the early stages know the risks - that there may be a good reason slugs run at this rate
Jun 30 00:39:04 <jacques> some slugs may become unstable
Jun 30 00:39:10 <jbowler-zzz> Yes, R83 is immediately below where R81 would be - so pretty accessible.
Jun 30 00:39:13 <jacques> `course I'm gonna do it tonight
Jun 30 00:39:13 <NAiL> I am willing to take that risk :)
Jun 30 00:39:26 <jacques> and start running torture tests - native builds
Jun 30 00:39:33 <rwhitby-away> should not mention on list at this time
Jun 30 00:39:43 <jacques> rwhitby-away, agreed
Jun 30 00:39:50 <NAiL> jacques: I've compiled for 50 mins or so. No problems so far
Jun 30 00:39:51 <rwhitby-away> also put a giant warning on wiki page
Jun 30 00:40:22 <jacques> NAiL, excellent - but you see my point right? we need a larger sample size before declaring this "somewhat safe"
Jun 30 00:40:33 <NAiL> yeah
Jun 30 00:40:40 <jacques> which is what we are implicitly doing the second we post to the masses
Jun 30 00:40:46 <rwhitby-away> maybe not mention on #nslu2-linux either
Jun 30 00:40:53 <NAiL> *cough*
Jun 30 00:41:13 <jacques> how many have been modded so far?
Jun 30 00:41:17 <rwhitby-away> oh, oops.
Jun 30 00:41:18 <NAiL> two, afaik
Jun 30 00:41:28 <jacques> both with R83 ?
Jun 30 00:41:35 <rwhitby-away> which resistor di beewoolie remove?
Jun 30 00:41:36 <jbowler-zzz> yes.
Jun 30 00:41:43 <jacques> beewoolie and NAiL ?
Jun 30 00:41:45 <NAiL> R64
Jun 30 00:41:48 <NAiL> both of us
Jun 30 00:41:50 <jbowler-zzz> beewoolie and NAiL both removed R64
Jun 30 00:41:57 <jacques> oops sorry, both with R64
Jun 30 00:42:00 <jbowler-zzz> (oops, sorry)
Jun 30 00:42:05 <NAiL> heh
Jun 30 00:42:28 <jbowler-zzz> Damn, just remove all of them then put them back 1 by 1 until it works.
Jun 30 00:42:43 <jacques> lol
Jun 30 00:42:45 <rwhitby-away> R83 must work
Jun 30 00:42:54 <jacques> we don't need no steenkin' resistors
Jun 30 00:43:00 <rwhitby-away> or we get our money back from intel
Jun 30 00:43:05 <NAiL> I don't get another slug to try with until later today or tomorrow
Jun 30 00:43:41 <jbowler-zzz> I want to know whether LinkSys did this intentionally or whether their screwy ordering of the resistors on the board caused an error.
Jun 30 00:43:58 <jacques> I say intentional
Jun 30 00:44:06 <NAiL> I'd say intentional too
Jun 30 00:44:08 <rwhitby-away> R83 looks easy to remove
Jun 30 00:44:14 <jacques> whoever did the board (not linksys probably)
Jun 30 00:44:21 <rwhitby-away> could do it with pliers ...
Jun 30 00:44:24 <NAiL> rwhitby-away: Yeah, shouldn't take more than a few secs
Jun 30 00:44:27 <jbowler-zzz> Then they must have talked to Intel, I guess Cisco carries some weight with Intel.
Jun 30 00:44:39 <jacques> heh I gess so :-)
Jun 30 00:44:53 <jacques> okay so which slug to mod.... I say the non-jtag one
Jun 30 00:45:20 <jbowler-zzz> I find it difficult to believe, they are using an undocumented setting for a non-obvious purpose - unless it does seriously overheat when it is in the box.
Jun 30 00:45:48 <jacques> I suspect heat or stability issues caused them to do it
Jun 30 00:46:05 <jacques> but I don't mind active cooling mine if I have to
Jun 30 00:46:13 <rwhitby-away> does it change the mem bus speed?
Jun 30 00:46:18 <jacques> partly because it would just be cool
Jun 30 00:46:36 <jacques> no pun intended
Jun 30 00:46:51 <NAiL> rwhitby-away: Any way to test?
Jun 30 00:47:00 <jbowler-zzz> mem speed: no, core clock only.
Jun 30 00:47:27 <jbowler-zzz> Memory speed is drive off the other registers I think (the chip select regs?)
Jun 30 00:47:40 <jacques> there are indeed ways to test
Jun 30 00:47:47 <jacques> hdparm, lmbench
Jun 30 00:47:55 <jbowler-zzz> SDRAM speed is really slow though - a lot slower than 133MHz.
Jun 30 00:47:55 <NAiL> jacques: hdparm?
Jun 30 00:47:57 <jacques> hdparm is the easiest
Jun 30 00:48:42 <NAiL> How do I test mem speed with hdparm?
Jun 30 00:48:46 <jbowler-zzz> To test the memory speed you need non-cached accesses - that's available in the physical address space.
Jun 30 00:49:33 <jacques> hdparm -T
Jun 30 00:49:42 <NAiL> ~200mb/sec
Jun 30 00:49:44 <jacques> just do a regular hdparm -Tt /dev/sda
Jun 30 00:49:48 <jacques> really?????
Jun 30 00:49:50 <jacques> no way
Jun 30 00:49:56 <NAiL> no
Jun 30 00:50:03 <NAiL>  Timing cached reads:   216 MB in  2.01 seconds = 107.46 MB/sec
Jun 30 00:50:07 <jacques> ah ok
Jun 30 00:50:08 <rwhitby-away> any other external speeds affected?
Jun 30 00:50:09 <jacques> that's same
Jun 30 00:50:51 <NAiL> Ok, what resistor do I have to desolder to OC the mem bus? :-P
Jun 30 00:50:57 <jacques> lol
Jun 30 00:51:03 <jacques> the ram probably can't handle it
Jun 30 00:51:24 <jacques> ram running at 1/2 proc is normal on xscale
Jun 30 00:51:47 <jacques> well on slower SA and xscale
Jun 30 00:52:05 <jacques> on my 400MHz pxa's it's running at 100MHz
Jun 30 00:52:30 <rwhitby-away> so the worst thing that can happen is the CPU runs hotter, and a passive heatsink might fix that
Jun 30 00:52:35 <NAiL> Started a perl config,make,make test. Wonder how long it'll take.
Jun 30 00:52:59 <jbowler-zzz> rwhitby-away: I believe everything coming out of the core is independently timed, and even the buses within the core (like the NPE ones) are independent.
Jun 30 00:53:14 <NAiL> The temperature on mine is about 40C. But that's rather inaccurate. Measuring with fingertips.
Jun 30 00:53:29 <NAiL> (after building perl for ~50 mins)
Jun 30 00:53:52 <rwhitby-away> in which case, unless Linksys got special chips from Intel, then we should be fine stability-wise as long as the CPU is adequately cooled.  Nothing else should be affected.
Jun 30 00:54:30 <jacques> neat my openslug powered down
Jun 30 00:54:38 <jacques> (on halt)
Jun 30 00:54:51 <NAiL> Mine always does that :D
Jun 30 00:56:02 <jacques> it dodn't work not too long ago
Jun 30 00:56:14 <rwhitby-away> let's discuss overclocking here only for the moment until we know it is safe
Jun 30 00:56:24 <rwhitby-away> i.e. only in #openjtag
Jun 30 00:56:49 <jbowler-zzz> Chapter 2.3 in the developer manual: the buses are hardwired to 133.32MHz and 66.66MHz (depending on bus).
Jun 30 00:57:22 <rwhitby-away> how about GPIO - will the I2C be affected?
Jun 30 00:57:40 <NAiL> rwhitby-away: That'd show with the leds, right?
Jun 30 00:58:24 <rwhitby-away> dunno
Jun 30 00:58:25 <jbowler-zzz> That's on the APB bus I think.
Jun 30 00:58:45 <NAiL> I'll run anything you care to throw at my slug ;)
Jun 30 00:59:32 <NAiL> *sigh*
Jun 30 00:59:32 <NAiL> Alignment trap: try (6355) PC=0x000085e0 Instr=0xe79e1000 Address=0xbefffdd9 FSR 0x013
Jun 30 00:59:36 <NAiL> Alignment trap: try (6355) PC=0x000085e0 Instr=0xe79e1000 Address=0xbefffdd9 FSR 0x013
Jun 30 00:59:39 <NAiL> Alignment trap: try (6355) PC=0x000085e0 Instr=0xe79e1000 Address=0xbefffdd9 FSR 0x013
Jun 30 00:59:42 <NAiL> Alignment trap: try (6355) PC=0x000085e0 Instr=0xe79e1000 Address=0xbefffdd9 FSR 0x013
Jun 30 00:59:45 <NAiL> *NOW* that perl bug struck me
Jun 30 01:00:20 <NAiL> How did I set the alignment?
Jun 30 01:00:32 <jbowler-zzz> That's another thing which is in the perl configure - alignment requirements (along there with byte sex).
Jun 30 01:00:55 <NAiL> The strange thing is that it didn't hit me before
Jun 30 01:01:15 <NAiL> Compiled perl natively a while back, and each time I only got one alignment trap
Jun 30 01:01:16 <jbowler-zzz> Did you change the CPU alignment?  (see /etc/init.d/alignment.sh)
Jun 30 01:01:43 <NAiL> no
Jun 30 01:02:25 <NAiL> It's at 1
Jun 30 01:02:36 <jbowler-zzz> Alignment issues are not reproducible if they originate in a storage allocator (can depend on allocation order).
Jun 30 01:05:39 <jacques> OK I am looking at a naked slug now
Jun 30 01:06:00 <jbowler-zzz> typical
Jun 30 01:06:19 <jacques> lol
Jun 30 01:06:22 <jacques> I see R64
Jun 30 01:06:28 <jacques> right by the flash chip
Jun 30 01:06:56 <jacques> and a strange black strip
Jun 30 01:07:34 <NAiL> R83 is on the pin-side of the flash. Right side, between the flash and the power button. 
Jun 30 01:07:43 <jbowler-zzz> R83 is to the right and below (at the board edge) - see the diagram next to R64
Jun 30 01:08:05 <jacques> yep see it now
Jun 30 01:08:15 <jacques> so which one to remove....
Jun 30 01:08:38 * NAiL cheers for R83
Jun 30 01:08:56 <rwhitby-away> R83 please
Jun 30 01:09:12 <NAiL> I'll be doing R83 on the slug I get next (unless you do it and it 'splodes)
Jun 30 01:09:38 <rwhitby-away> R83 is documented, near the edge, and has less important tracks around it to scratch
Jun 30 01:09:41 <jacques> R83 is right next to an unpopulated R81 space
Jun 30 01:09:52 <jacques> yeah, so R83 it is
Jun 30 01:09:52 <ByronT-Away> yep
Jun 30 01:10:18 <jacques> first must get coke to raise blood sugar
Jun 30 01:10:37 <rwhitby-away> no tracks under R83, so you could even use a chisel
Jun 30 01:10:43 <ByronT-Away> and caffeine to steady the hands!
Jun 30 01:10:47 <rwhitby-away> or a screwdriver
Jun 30 01:11:12 <rwhitby-away> whereas R64 has a fine track running right next to it
Jun 30 01:11:46 <rwhitby-away> which could be easily shorted to the large ground track next to it
Jun 30 01:17:08 * NAiL holds his breath
Jun 30 01:17:15 <jacques> soldering station setup
Jun 30 01:20:31 <jacques> needsomething to pry the resistor while I heat the solder
Jun 30 01:21:28 <NAiL> I used the iron itself. Came loose very easy
Jun 30 01:21:42 <jacques> ok
Jun 30 01:22:05 <rwhitby-away> jacques: there's nothing underneath it, so you can't hurt anything :-)
Jun 30 01:22:29 <peteru-home> Ask the wife to hand over the tweezers ;)
Jun 30 01:22:31 <rwhitby-away> and the flash chip will wick away the heat before it gets near the processor :-)
Jun 30 01:22:34 <jacques> it's off
Jun 30 01:23:01 <rwhitby-away> said the vicar to the ...
Jun 30 01:23:55 <jacques> ok iron off
Jun 30 01:26:21 <NAiL> ...
Jun 30 01:27:25 <rwhitby-treo> ...
Jun 30 01:27:30 <jacques> give me a few minutes to hook it to this machine to boot it
Jun 30 01:27:49 <jacques> gotta rehook the serial connection
Jun 30 01:28:30 * rwhitby-treo can just imagine what kind of cathartic release this is for jacques ...
Jun 30 01:30:53 <NAiL> Hmm.. what happens if I replace the 33Mhz clock with a 57Mhz I've got lying around? :P
Jun 30 01:30:54 <rwhitby-treo> ep1220: just in time
Jun 30 01:31:23 <ep1220> just in time for what ?
Jun 30 01:31:32 <NAiL> 266Mhz slug
Jun 30 01:31:37 <rwhitby-treo> jacques has just eemove R83
Jun 30 01:31:44 <rwhitby-treo> removed
Jun 30 01:31:59 <rwhitby-treo> as you suggested earlier today
Jun 30 01:32:26 <ep1220> Great result.
Jun 30 01:32:43 <NAiL> jacques' slug isn't up yet
Jun 30 01:33:01 <ep1220> bewoolie mailed me last nght: he said jaques removed R64 ??
Jun 30 01:33:14 <NAiL> bewoolie and I removed R64
Jun 30 01:33:19 <NAiL> Both our slugs are 266Mhz
Jun 30 01:33:43 <ep1220> Formula1 slug ;-)
Jun 30 01:33:46 <rwhitby-treo> R64 is undocumented and tested. R83 is documented and just about to be confirmed
Jun 30 01:34:01 <jacques> kernel 2.6 changed around the usb serial dev
Jun 30 01:34:26 <rwhitby-treo> F1Slug
Jun 30 01:34:47 <ep1220> R83 goes to which address line ?
Jun 30 01:34:58 <NAiL> 05:49 < dyoung-web> A22 = R81 A21 = R82 A23 = R64
Jun 30 01:35:01 <rwhitby-treo> gives 011
Jun 30 01:35:16 <jacques> beep
Jun 30 01:35:20 <NAiL> :D
Jun 30 01:35:23 <rwhitby-treo> documented 266 mode
Jun 30 01:35:38 <jacques> unfortunately this slug is running 2.6.12
Jun 30 01:35:43 <jacques> so I see no kernel boot messages
Jun 30 01:35:52 <rwhitby-treo> dmesg it son
Jun 30 01:35:57 <NAiL> use the dmesg
Jun 30 01:36:03 <NAiL> , luke
Jun 30 01:36:03 <jacques> it's not up yet, it just beeped
Jun 30 01:36:15 <jacques> flashing yellw now
Jun 30 01:36:26 <jacques> it's up to a prompt
Jun 30 01:36:41 <rwhitby-treo> did it seem faster :-)
Jun 30 01:36:41 <jbowler-zzz> NAiL: is that correct?  A21=R82?
Jun 30 01:36:44 <jacques> root@LKG0FB08F:~# cat /proc/cpuinfo 
Jun 30 01:36:44 <jacques> Processor       : XScale-IXP42x Family rev 1 (v5b)
Jun 30 01:36:44 <jacques> BogoMIPS        : 266.24
Jun 30 01:37:00 <NAiL> jbowler-zzz: According to dyoung. Hang on 2 sec.
Jun 30 01:37:29 <rwhitby-treo> what NAiL posted can't be right
Jun 30 01:37:30 <jbowler-zzz> jacques just removed R83
Jun 30 01:37:44 <NAiL> rwhitby-treo: Quite possible...
Jun 30 01:38:05 <NAiL> Ignore my paste ;)
Jun 30 01:38:21 <rwhitby-treo> NAiL: can you update the wiki page with R83?
Jun 30 01:38:27 <NAiL> yup, will do
Jun 30 01:39:03 <rwhitby-treo> can someone do a crop and circle from the PCB photo for that page?
Jun 30 01:39:21 <rwhitby-treo> to show R83 location?
Jun 30 01:39:47 <rwhitby-treo> we also should put a donation button or two on that page ... ;-)
Jun 30 01:40:03 <jacques> this is such a vindication
Jun 30 01:40:05 <rwhitby-treo> I expect it will get a few hits ...
Jun 30 01:40:21 <rwhitby-treo> we may even get our first real slashdot ...
Jun 30 01:40:37 <rwhitby-treo> (we only got a sixth order one before)
Jun 30 01:41:10 <NAiL> Wiki updated.
Jun 30 01:41:32 <jacques> I think we're more likely to get slashdotted if we mount a fan  :-D
Jun 30 01:41:38 <NAiL> haha
Jun 30 01:42:13 <rwhitby-treo> casefancase?
Jun 30 01:42:19 <jacques> lol
Jun 30 01:42:28 <jacques> i would say water cooling but that seems a bit overkill
Jun 30 01:42:32 <NAiL> haha
Jun 30 01:42:43 <NAiL> hmm..
Jun 30 01:42:46 * NAiL considers the thought
Jun 30 01:42:46 <rwhitby-treo> slime cooling
Jun 30 01:42:53 <NAiL> I *do* have a water cooling system
Jun 30 01:43:00 <NAiL> I could just slap the slug onto that
Jun 30 01:43:04 <jacques> lol
Jun 30 01:43:10 <NAiL> That'd get /.'s attention! :D
Jun 30 01:43:16 <jacques> no I know - we immerse the slug in mineral oil
Jun 30 01:43:20 <jacques> they love that 
Jun 30 01:43:27 <jacques> (on /.)
Jun 30 01:43:34 <NAiL> or that intert-thingy-fluid
Jun 30 01:43:46 <NAiL> fluorinert
Jun 30 01:43:58 <jacques> too $$$
Jun 30 01:44:05 <rwhitby-treo> hmm - ka6sox gets slashdotted while in Albania ...
Jun 30 01:44:05 <jacques> mineral oil works fine
Jun 30 01:44:14 <rwhitby-treo> ouch
Jun 30 01:44:48 <jacques> very good point rwhitby-
Jun 30 01:44:53 <jacques> treo
Jun 30 01:44:58 <NAiL> yup
Jun 30 01:45:56 <rwhitby-treo> I could put up a overclock.nslu2-linux.org vhost on sf.net ...
Jun 30 01:46:19 <jacques> BTW, if ppl running overclocked slugs see a process called "supersoldier" just ignore it
Jun 30 01:46:32 <NAiL> lol
Jun 30 01:46:41 <rwhitby-treo> sshhh!
Jun 30 01:46:52 <jacques> actually, download my enhanced init binary and it'll "go away"
Jun 30 01:47:19 <rwhitby-treo> SlugArmyOnSpeed
Jun 30 01:47:36 <jacques> better than on crack I guess :-)
Jun 30 01:47:36 <NAiL> I've got a 10GB/mo mirror. D'ya think /. will kill that? :)
Jun 30 01:47:59 <rwhitby-treo> yep
Jun 30 01:48:21 <rwhitby-treo> nah :-)
Jun 30 01:48:24 <jacques> this may sound a bit mercenary, but if we think we're going to get /.ed maybe we should try to take advantage of the publicity
Jun 30 01:48:35 <rwhitby-treo> yep
Jun 30 01:48:39 * NAiL agrees
Jun 30 01:48:46 <NAiL> Plaster the page with donate buttons
Jun 30 01:48:47 <jacques> jacques needs a new pair of boots
Jun 30 01:48:56 <rwhitby-treo> you have to click through paypal to get the info ...
Jun 30 01:49:04 <NAiL> make "onmouseover" events that opens paypal
Jun 30 01:49:21 <NAiL> Make 'em pay to see the pictures
Jun 30 01:49:42 <NAiL> Hey, let's affiliate with someone who sells nslu2's
Jun 30 01:49:52 <NAiL> They'll probably get a lot of customers ;)
Jun 30 01:50:01 <jacques> the list of mods gets a little longer tonight
Jun 30 01:50:10 <rwhitby-treo> amazon click-throughs ...
Jun 30 01:50:26 <NAiL> google ads ;)
Jun 30 01:50:26 <jacques> how much would you pay for a 64MB serial jtaged overclocked slug? :-)
Jun 30 01:50:38 <jacques> with heatsink!
Jun 30 01:50:50 <NAiL> not more than I'd pay for one that wasn't overclocked :-P
Jun 30 01:51:11 <jacques> not talking about ppl who can mod their own
Jun 30 01:51:16 <rwhitby-treo> wrong group of people to ask
Jun 30 01:51:19 <jacques> can/want to
Jun 30 01:51:25 <jacques> yeah I realize that now
Jun 30 01:51:36 <jacques> tho I cannot do the memory upgrade myself (yet)
Jun 30 01:51:48 <jacques> and where the heck is dyoung ?
Jun 30 01:52:06 <jacques> he should be here :-(
Jun 30 01:52:12 <rwhitby-treo> yeah
Jun 30 01:52:21 <rwhitby-treo> he will kick himself
Jun 30 01:52:32 <jacques> heck I almost missed it myself :-)
Jun 30 01:52:35 <jacques> damn sleep
Jun 30 01:52:53 <NAiL> I was supposed to go to sleep long ago
Jun 30 01:53:00 <jacques> heh
Jun 30 01:53:05 <NAiL> 20 secs after I said good night, bewoolie said WAIT!
Jun 30 01:53:58 <NAiL> That was more than three hours ago
Jun 30 01:54:20 <jacques> NAiL, been there :-)
Jun 30 01:54:33 <rwhitby-treo> goodbye BK, hello SpeedySlug
Jun 30 01:55:15 <NAiL> Good :D
Jun 30 01:55:32 <jacques> mine's called SlugRacer®  ;-P
Jun 30 01:56:02 <NAiL> Mine's still Maimedslug. Albeit I'm considering Maimedslug^2
Jun 30 01:57:27 <rwhitby-treo> ok, looks like I need to do some soldering tonight ...
Jun 30 01:57:43 <NAiL> I'm hoping I get my new slug today :)
Jun 30 01:57:48 <rwhitby-treo> devslug -> speedyslug
Jun 30 01:57:50 <NAiL> If not, I'll solder tomorrow
Jun 30 01:58:22 <rwhitby-treo> so any effect on disk serving performance?
Jun 30 01:58:26 <NAiL> nope
Jun 30 01:58:29 <jacques> i sure wish ixp's had an internal temperature sensor
Jun 30 01:58:37 <rwhitby-treo> cause if not, it won't sell
Jun 30 01:58:40 <NAiL> I get ~6.2mb sec from my hdd
Jun 30 01:58:56 <jacques> I need to reassemble mine and reattach net and hd
Jun 30 01:58:59 <rwhitby-treo> not hdparm, samba serving with stockish firmware
Jun 30 01:59:10 <jacques> it's just running naked on  apaper towel right now
Jun 30 01:59:11 <rwhitby-treo> joe user impact
Jun 30 01:59:18 <NAiL> No idea...
Jun 30 01:59:34 <NAiL> I've never run stockish firmware :P
Jun 30 01:59:54 <jacques> rwhitby-treo, I'll run a full set of tests (others probably are too)
Jun 30 02:00:04 <NAiL> Mine's building perl
Jun 30 02:00:06 <rwhitby-treo> jacques - can you load speedyslug with unslung?
Jun 30 02:00:12 <jacques> id rather now
Jun 30 02:00:13 <jacques> not
Jun 30 02:00:15 <jacques> why?
Jun 30 02:00:29 <NAiL> rwhitby-treo: If I get my newslug later today, I'll load that one up with unslung
Jun 30 02:00:30 <jacques> you think unslung kernel may have issues with it?
Jun 30 02:00:47 <rwhitby-treo> cause that is the advertised test - speed of samba with stock firmware
Jun 30 02:00:57 <jacques> oh
Jun 30 02:01:02 <jacques> what's samba?
Jun 30 02:01:08 <jacques> :-)
Jun 30 02:01:09 <rwhitby-treo> hardware mod is independent of custom firmware ...
Jun 30 02:01:16 <jacques> sounds like some windows thing
Jun 30 02:01:32 <ep1220> jacques: Did you remove R83 only, or R64 too ?
Jun 30 02:01:40 <jacques> on;y R83
Jun 30 02:01:46 <jacques> only R83
Jun 30 02:01:47 <rwhitby-treo> R83 only - 011
Jun 30 02:01:54 <rwhitby-treo> documented 266
Jun 30 02:01:55 <NAiL> rwhitby-treo: I get ~4.5mb/sec via ftp. I suspect samba isn't going to be much better :(
Jun 30 02:02:19 <jacques> remember ram is still at same speed
Jun 30 02:02:30 <jacques> andsince it's not 1:1 anymore, there are wait states involved
Jun 30 02:02:35 <rwhitby-treo> if it brings the samba perf closer to ftp perf then that is significant
Jun 30 02:02:57 <rwhitby-treo> think joe user ...
Jun 30 02:03:07 <NAiL> Uh... I got more than 5mb with samba on openslug earlier
Jun 30 02:03:25 <rwhitby-treo> NAiL: think joe user
Jun 30 02:03:51 * NAiL thinks
Jun 30 02:04:06 <rwhitby-treo> unmodified slug, open case, break resistor, what increase in Linksys stock functionality performance?
Jun 30 02:04:12 <jacques> joe user as in unslung ?
Jun 30 02:04:18 <rwhitby-treo> stock
Jun 30 02:04:33 <rwhitby-treo> (equiv to testing with unslung)
Jun 30 02:04:47 <jacques> I'm going to remove all the resistors now and see how fast it goes.
Jun 30 02:04:51 <rwhitby-treo> joe user as in not even custom firmware
Jun 30 02:04:58 <rwhitby-treo> it won't go faster
Jun 30 02:05:15 <rwhitby-treo> fuses inside chip
Jun 30 02:05:36 <rwhitby-treo> limit top speed
Jun 30 02:05:53 * NAiL curses at debian
Jun 30 02:13:39 <NAiL> I keep refreshing the package tracing window. Hoping that the slug arrives faster ;)
Jun 30 02:13:49 <rwhitby-away> heh
Jun 30 02:23:56 <jacques> it might be my imagination, but it seems like the "uncompressing kernel..." is much faster now
Jun 30 02:24:20 <NAiL> Then it is my imagination as well ;)
Jun 30 02:24:35 <rwhitby-away> it should be faster
Jun 30 02:25:54 <NAiL> bleh... wish that package tracing had GPS
Jun 30 02:26:41 <NAiL> I know that the slug will drive by just outside my window many hours before I get it
Jun 30 02:29:30 <jacques> it might be my imagination, but openssh seems faster now
Jun 30 02:29:49 <NAiL> hehe, dropbear is definitely faster
Jun 30 02:29:52 <jacques> oops this one is running dropbear
Jun 30 02:30:39 <NAiL> I don't have to wait two secs for dropbear to accept a connection anymore
Jun 30 02:30:48 <jacques> yeah
Jun 30 02:32:09 <NAiL> hmm
Jun 30 02:32:16 <NAiL> One thing with this overclocking..
Jun 30 02:32:29 <NAiL> How much more power will the slug draw?
Jun 30 02:32:52 <jacques> someone with the correct equipment needs to check
Jun 30 02:33:49 <ep1220> I'd say: The CPU will roughly double its power consumption
Jun 30 02:34:24 <ep1220> the rest of the system still running at same clock, so will not change.
Jun 30 02:35:49 <NAiL> dyoung: here :)
Jun 30 02:36:27 <NAiL> dyoung: There's three slugs running at 266. One with R83 removed (according to data sheet), and two with R64 removed. *Seems* to have same effect.
Jun 30 02:37:13 <rwhitby-away> I propose R83 be the official mod
Jun 30 02:37:34 * NAiL agrees (Although I'm not "official") ;)
Jun 30 02:37:39 <jacques> I think the first person to do the official mod should get a prize
Jun 30 02:37:44 <NAiL> heh
Jun 30 02:37:49 <jacques> ;-P
Jun 30 02:37:52 <rwhitby-away> was that beewoolie?
Jun 30 02:38:03 <NAiL> no, jacques
Jun 30 02:38:06 <jacques> nope, hmm it seems it was me :-D
Jun 30 02:38:13 <rwhitby-away> oh, right :-)
Jun 30 02:38:13 <NAiL> beewolie and I did the unofficial one ;)
Jun 30 02:38:16 <jacques> what a coinkidink
Jun 30 02:38:22 <NAiL> lol
Jun 30 02:38:42 <rwhitby-away> jacques: you get to be the OpenSlug native package feed manager - how's that for a prize!
Jun 30 02:38:50 <NAiL> Looks like I won't be getting the newslug until tomorrow. :(
Jun 30 02:38:57 <jacques> yay!!  uh... ?  :-)
Jun 30 02:39:13 <jacques> I was hoping for new boots :-(
Jun 30 02:39:27 <NAiL> close, but no cigar
Jun 30 02:39:49 <NAiL> That's for the second modder, unofficial or not :-P
Jun 30 02:39:53 <jacques> or maybe a nice nokia 770
Jun 30 02:40:27 <dyoung> Ok
Jun 30 02:40:29 <dyoung> that does it
Jun 30 02:40:31 <dyoung> theres too much log.
Jun 30 02:40:35 <dyoung> Somone just spit it out.
Jun 30 02:40:36 <NAiL> hahaha
Jun 30 02:40:39 <NAiL> It works.
Jun 30 02:40:39 <dyoung> Does it work?
Jun 30 02:40:48 <dyoung> Is it hotter?
Jun 30 02:40:52 <NAiL> I've been compiling perl on it for quite some time now
Jun 30 02:41:05 <NAiL> It runs at about 40C, I'd say
Jun 30 02:41:06 <dyoung> And dammit, I'm gonna be like 10th on the list by the time I get my stuff together.
Jun 30 02:41:06 <jacques> we need a way to measure the temps
Jun 30 02:41:13 <NAiL> That's measurement by finger
Jun 30 02:41:29 <jacques> NAiL, but what was it before?
Jun 30 02:41:31 <dyoung> what is the preferred method of doing this?  Remove R64?  R83?  or ADD a new one?
Jun 30 02:41:34 <NAiL> Perl is now doing tests
Jun 30 02:41:41 <NAiL> dyoung: Remove R83
Jun 30 02:41:51 <dyoung> R83 is a chip select though.
Jun 30 02:41:58 <jacques> dyoung, official method is remove R83
Jun 30 02:41:59 <NAiL> 'http://www.nslu2-linux.org/wiki/HowTo/OverClockTheSlug
Jun 30 02:42:07 <dyoung> Dammit, I missed it!@
Jun 30 02:42:11 <dyoung> Now I'm mad!@
Jun 30 02:42:49 <rwhitby-away> told you he'd be mad :-)
Jun 30 02:42:55 <NAiL> haha
Jun 30 02:42:57 <jacques> uh-oh Angry Dyoung
Jun 30 02:43:03 <rwhitby-away> ADS
Jun 30 02:43:18 <rwhitby-away> s/Woman/DYoung/
Jun 30 02:43:37 <jacques> holy craap
Jun 30 02:43:46 <jacques> I just timed perl Configure -de
Jun 30 02:43:57 <rwhitby-away> and ...
Jun 30 02:43:59 <jacques> previous value: real    18m53.431s
Jun 30 02:44:06 <jacques> now: real    10m7.199s
Jun 30 02:44:12 <rwhitby-away> woof!
Jun 30 02:44:12 <dyoung> sigh
Jun 30 02:44:15 <jacques> now to time the make 
Jun 30 02:44:20 <dyoung> I cant believe I missed it.
Jun 30 02:44:33 <NAiL> jacques: I'm timing config,make,maketest
Jun 30 02:44:35 <dyoung> oh well, at least I could spit it out before getting beaten up.
Jun 30 02:44:54 <rwhitby-away> dyoung: you were here in spirit - you'll see we channelled your spirit at the special moment ....
Jun 30 02:45:00 <jacques> yep
Jun 30 02:45:33 <NAiL> lib/Memoize/t/speed.......................FAILED at test 2
Jun 30 02:45:38 <NAiL> That one's new...
Jun 30 02:45:42 <ep1220> my clock-test prog draws 35mA at 133MHz; but is no ideal test, i guess
Jun 30 02:46:05 <NAiL> So it'll probably draw around 70mA at 266
Jun 30 02:46:07 <NAiL> ?
Jun 30 02:46:07 <jacques> NAiL, hmm
Jun 30 02:46:18 <ep1220> NAiL: no
Jun 30 02:46:22 <NAiL> jacques: Might be me choosing the wrong alignment-thingy
Jun 30 02:46:34 <NAiL> oh, that's the *whole* slug?
Jun 30 02:46:39 <ep1220> The CPU is only a fraction of what consumes power onthe board
Jun 30 02:46:42 <jacques> we'll compare results when mine's done
Jun 30 02:47:06 <ep1220> Are you aware of a small CPU-stress prog ?
Jun 30 02:47:35 <NAiL> ep1220: You can always loop the e-finding program we used to bench
Jun 30 02:47:39 <dyoung> Ok, theres just too much log for me right now.
Jun 30 02:47:41 <rwhitby-away> I should have one ready in an hour or so ..
Jun 30 02:47:47 <rwhitby-away> (a SpeedySlug that is)
Jun 30 02:47:51 <dyoung> so thats remove R83 ONLY?  or move R83 to R81 ?
Jun 30 02:48:04 <dyoung> Or remove R64 Only?
Jun 30 02:48:10 <NAiL> remove 83 only
Jun 30 02:48:12 <NAiL> that's it
Jun 30 02:48:19 <NAiL> no moving, no removing anything else
Jun 30 02:48:33 <NAiL> heat it a little bit and push it away :)
Jun 30 02:49:10 <jacques> lol
Jun 30 02:49:19 <jacques> my perl build failed on toke.c
Jun 30 02:49:24 <jacques> I had forgotten to enable swap
Jun 30 02:50:25 <NAiL> ep1220: try looping this: http://pastebin.ca/16462 
Jun 30 02:51:02 <rwhitby-away> bbiab
Jun 30 02:51:20 <rwhitby-away> remove R83 only.  nothing else
Jun 30 02:51:39 <rwhitby-away> Do we need a SpeedySlug DB on yahoo ?
Jun 30 02:51:53 <NAiL> Doesn't hurt
Jun 30 02:52:01 <dyoung> people will wonder what that is
Jun 30 02:52:06 <dyoung> do we want to let this cat out of the bag yet?
Jun 30 02:52:31 <dyoung> I really am cranky that I missed all the fun BTW.
Jun 30 02:52:54 <NAiL> I was waiting for you all the time!
Jun 30 02:53:58 <peteru-home> TurboSlug should help with puppy performance. rwhitby-away should collect some stats for before and after :-)
Jun 30 02:54:23 <dyoung> sigh
Jun 30 02:54:31 <dyoung> oh well, at least it seems to work.
Jun 30 02:54:44 <NAiL> Configuring, building and maketesting perl took:
Jun 30 02:54:44 <NAiL> real    120m20.740s
Jun 30 02:54:45 <NAiL> user    102m16.520s
Jun 30 02:54:45 <NAiL> sys     12m33.560s
Jun 30 02:55:04 <NAiL> That includes the 5-10 minutes with assertion failures
Jun 30 02:55:26 <jacques> previous values were 19+90+75
Jun 30 02:55:43 <NAiL> Not as much improvement as I had hoped
Jun 30 02:55:54 <jacques> uh
Jun 30 02:55:59 <NAiL> Shaves off an hour, approx
Jun 30 02:56:17 <jacques> how much improvement did you think you'd get ??
Jun 30 02:56:22 <jacques> it's almost double speed
Jun 30 02:56:36 <jacques> (accounting for the assertion thing)
Jun 30 02:56:55 <NAiL> I expected a 10-15 minutes less ;)
Jun 30 02:57:35 <NAiL> Originally it took 3 hours 5 mins. Now it took 1 hour 50.
Jun 30 02:59:14 <jacques> I'll compare the compile-only times
Jun 30 02:59:31 <jacques> I suspect the tests might have not seen as much improvement 
Jun 30 02:59:52 <jacques> it depends on the workload - how muc ram access/disk access is involved
Jun 30 03:00:24 <jacques> well, time for me to make breakfast
Jun 30 03:00:35 <jacques> I put off eating for this :-)
Jun 30 03:00:59 <dyoung> so do the "offending" parties at least get some gold stars?
Jun 30 03:01:25 <jacques> offending?
Jun 30 03:01:51 <dyoung> "offending" as in those who caused this discovery.
Jun 30 03:02:15 <jacques> yep :-)
Jun 30 03:02:24 <jacques> I need a terminal in my kitchen
Jun 30 03:02:51 <jacques> so I can use screen and watch what's going on while cooking
Jun 30 03:03:20 <dyoung> so who all has done this mod already?
Jun 30 03:03:58 <jacques> beewollie and NAiL did R64, jacques did R83
Jun 30 03:04:01 <jacques> that's it
Jun 30 03:04:57 <dyoung> ok so if I hurry I can be number 4.
Jun 30 03:05:54 <NAiL> You've gotta race rwhitby ;)
Jun 30 03:07:52 <NAiL> Doesn't seem like it gets any hotter while running at 100% cpu
Jun 30 03:08:12 <NAiL> Still around the low 40C's
Jun 30 03:09:39 <jacques> do we fully understand what's going on? why this works?
Jun 30 03:10:15 <rwhitby-away> yes, and it's documented
Jun 30 03:10:22 * rwhitby-away is getting out the soldering iron now
Jun 30 03:10:23 <dyoung> I'm still working on why R83 works
Jun 30 03:10:33 <rwhitby-away> dyoung: 011
Jun 30 03:10:44 <dyoung> But R83 isnt one of the lines in question?
Jun 30 03:10:51 <dyoung> Its R64 R81 R82 ?
Jun 30 03:10:56 <rwhitby-away> I thought it was the '0' in 011 ...
Jun 30 03:11:40 <dyoung> 	<dyoung-web>	A22 = R81 A21 = R82 A23 = R64
Jun 30 03:11:52 <rwhitby-away> hmm - do I populate the new monotone db, or do I speed up three slugs ....
Jun 30 03:12:15 <NAiL> rwhitby-away: Go speed up slugs, before dyoung! ;-)
Jun 30 03:12:30 <dyoung> Well, before I go and remove R83 I wanna know why it works
Jun 30 03:12:42 <dyoung> because right now I dont have any clue
Jun 30 03:13:05 <dyoung> R83 is hooked to a chip select.
Jun 30 03:13:41 <rwhitby-away> are you telling us it should be R81 instead?
Jun 30 03:14:01 * rwhitby-away goes to work on monotone until dyoung sorts it out ...
Jun 30 03:15:47 <NAiL> This is one exciting day. BK dies, Slugs start running fast. 
Jun 30 03:15:50 <NAiL> What's next?
Jun 30 03:16:06 <rwhitby-away> dyoung: what's pin 3 on the flash?
Jun 30 03:16:29 <dyoung> A21 I believe.
Jun 30 03:16:41 <rwhitby-away> that's what R83 is connected to
Jun 30 03:16:45 <NAiL> Then A21 -> R83
Jun 30 03:17:57 <dyoung> hahahahaha
Jun 30 03:17:58 <dyoung> okay
Jun 30 03:18:17 <dyoung> I must have really been under the fear of ass-kickage.
Jun 30 03:19:28 <dyoung> Slugs apart, iron is hot
Jun 30 03:20:07 <jacques> good time to strike then
Jun 30 03:21:01 <dyoung> lifted and booting
Jun 30 03:21:23 <dyoung> Wow, it is faster.
Jun 30 03:21:32 <NAiL> Congratulations. You're #4!
Jun 30 03:21:40 <jacques> rwhitby-away, I'm not sure I would do *all* my slugs quite yet
Jun 30 03:21:55 <rwhitby-away> jacques: ah, but you're not me :-)
Jun 30 03:22:00 <jacques> lol
Jun 30 03:22:10 <jacques> do we now the value of those resistors?
Jun 30 03:22:14 <jacques> know
Jun 30 03:22:19 <dyoung> 10k
Jun 30 03:22:22 <jacques> ah
Jun 30 03:22:30 <dyoung> 103 = 10 * 10^3
Jun 30 03:22:36 * jacques thinks about a slug turbo switch.
Jun 30 03:22:48 <rwhitby-away> it's an address line, you don't want a switch
Jun 30 03:23:10 <dyoung> Beewoolie earlier pointed out that extending the line may introduce other problems because of the length of wire.
Jun 30 03:23:30 <dyoung> if its a few mm it might be okay
Jun 30 03:23:32 <jacques> i thought it was only read on chip reset
Jun 30 03:23:44 <jacques> but I probably confused myself readong that appnote
Jun 30 03:24:07 <dyoung> Thats when its read.  the problems would be during the high speed transfers during normal operation
Jun 30 03:24:20 <jacques> oh, ok
Jun 30 03:25:30 <dyoung> sorry for the misinformation about R81.
Jun 30 03:25:41 <dyoung> (not like it slowed anyone down....)
Jun 30 03:26:02 <dyoung> I only have 263 BogoMips.
Jun 30 03:26:08 <dyoung> Where are my other 3 BogoMips?
Jun 30 03:26:11 <dyoung> I feel ripped off.
Jun 30 03:26:22 <NAiL> You're not in the 266BogoMips club!
Jun 30 03:26:27 <NAiL> Booo-hisss
Jun 30 03:27:01 <dyoung> rebooting
Jun 30 03:28:13 <dyoung> Still 263.78 BogoMips.
Jun 30 03:28:18 <peteru-home> dyoung: perhaps you left a blob of solder on the trace and it's big enough to absorb 3 BogoMIPSen ;-)
Jun 30 03:28:34 <NAiL> Odd clock?
Jun 30 03:28:36 <dyoung> I hit it with the thermometer.  Its 41degC
Jun 30 03:28:41 <rwhitby-away> resistor removed
Jun 30 03:28:55 <dyoung> I didnt remove the resistor actually, I jus tlifted one side.
Jun 30 03:29:01 <rwhitby-away> booting
Jun 30 03:29:09 <dyoung> maybe 3Bogomips are leaking through. :-)
Jun 30 03:29:50 <dyoung> jbowler-zzz, Thanks a LOT for those two lines of code earlier.
Jun 30 03:30:06 <dyoung> I'll thank you again when youre not zzz
Jun 30 03:31:08 <jacques> dyoung, which exact kernel you running?
Jun 30 03:31:29 <jacques> that can affect bogomips by small amounts 
Jun 30 03:31:31 <dyoung> we have a technician onsite in a place literally around the world from here that was suffering because at the exact moment he started messing with stuff, the downlink from the satelite went down.  so we had to get the data through alternate means.
Jun 30 03:31:51 <dyoung> jacques, Unslung 2.4
Jun 30 03:31:58 <NAiL> jacques: I'm running 2.6.11-2, and get 266.24
Jun 30 03:32:00 <dyoung> is what h appened to be loaded in flash
Jun 30 03:32:01 <jacques> dyoung, ah, I'm on 2.6.12
Jun 30 03:32:13 <jacques> NAiL, yeah, I get same as you
Jun 30 03:32:23 <dyoung> okay so unslung steals 3 BogoMips
Jun 30 03:32:38 <dyoung> 43Deg.
Jun 30 03:32:40 <dyoung> Idle
Jun 30 03:32:45 <NAiL> wow
Jun 30 03:33:01 <NAiL> I'm pretty sure yours is running hotter than mine
Jun 30 03:33:04 <jacques> I've seen processors run fine above 80C
Jun 30 03:33:07 <dyoung> Someone have a binary I can throw at it to make it spin for a while?
Jun 30 03:33:21 <NAiL> Do you have native compile?
Jun 30 03:33:25 <dyoung> No
Jun 30 03:33:32 <dyoung> I have a unslung slug with no drives.
Jun 30 03:33:36 <jacques> dyoung, yes ?
Jun 30 03:33:37 <NAiL> heh, ok
Jun 30 03:33:49 <jacques> that's what I use to peg cpu sometimes
Jun 30 03:33:58 <dyoung> jacques, will your binary work under 2.4 ok?
Jun 30 03:34:05 <NAiL> "yes"
Jun 30 03:34:17 <jacques> who's on first?
Jun 30 03:34:30 <jacques> yes > /dev/null
Jun 30 03:34:38 <NAiL> /usr/bin/yes
Jun 30 03:34:50 <dyoung> I dont have yes
Jun 30 03:34:54 <jacques> dang
Jun 30 03:35:09 <dyoung> cat /dev/zero > /dev/null ?
Jun 30 03:35:10 <jacques> is it built into busybox and just not symlinked ?
Jun 30 03:35:25 <dyoung> no yes in that busybox
Jun 30 03:35:31 <jacques> :-\
Jun 30 03:35:32 <NAiL> heh
Jun 30 03:35:45 <dyoung> slingbox...
Jun 30 03:35:47 <dyoung> checking
Jun 30 03:35:58 <dyoung> no yes there either.
Jun 30 03:36:12 <dyoung> will my cat make the CPU spin?
Jun 30 03:36:20 <NAiL> cat /dev/zero | gzip - |gunzip - > /dev/null
Jun 30 03:36:28 <NAiL> you have gzip/gunzip, right?
Jun 30 03:36:31 <dyoung> yes
Jun 30 03:36:32 <rwhitby-away> Hmm - devslug don't get past the "Executing boot script" line ...
Jun 30 03:36:56 <NAiL> That'll eat cpu :)
Jun 30 03:37:14 <dyoung> okay running
Jun 30 03:37:22 <jacques> there needs to be a cpuburn for arm
Jun 30 03:37:33 <rwhitby-away> dyoung: mine's not :-(
Jun 30 03:37:34 <NAiL> I've been running at 100% for 20 mins. Still in the lower end of the 40s. 
Jun 30 03:37:34 <jacques> cpuburn is really useful when testing x86
Jun 30 03:37:44 <jacques> rwhitby-away, :-\
Jun 30 03:37:46 <dyoung> 46deg
Jun 30 03:38:06 <NAiL> 46 is like... hot
Jun 30 03:38:08 <dyoung> rwhitby-away, comfort.
Jun 30 03:38:23 <dyoung> does replacing it help?
Jun 30 03:38:55 <jacques> NAiL, I've seen my t42 get above 80C under load at full speed
Jun 30 03:39:02 <jacques> when it's hot in here
Jun 30 03:39:04 <NAiL> t42?
Jun 30 03:39:11 <jacques> laptop
Jun 30 03:39:13 <NAiL> ah
Jun 30 03:39:19 <dyoung> nail, what is your ambient air temp?
Jun 30 03:39:29 <NAiL> 25-26 atm
Jun 30 03:39:36 <dyoung> ok, mine is 29
Jun 30 03:39:42 <dyoung> so we're in the same ballpark
Jun 30 03:39:50 <dyoung> (of CPU temp)
Jun 30 03:39:53 <NAiL> You're running without case, I presume?
Jun 30 03:40:33 <peteru-home> rwhitby-away: Lucky you got two more slugs to sacrifice :-/
Jun 30 03:40:34 <dyoung> yeah, and hitting it with this infrared thermometer for temp reading
Jun 30 03:41:19 <dyoung> 47Deg
Jun 30 03:41:38 <NAiL> Hmm.. I don't like your readings
Jun 30 03:41:57 <NAiL> You're actually starting to get up to the same level as my AMD2700+
Jun 30 03:42:17 <dyoung> its onjly 47 4mm to the rght of the "i"
Jun 30 03:42:23 <dyoung> everywhere else is 44
Jun 30 03:42:24 <rwhitby-away> peteru-home: second one done, and working 266.24 BogoMips
Jun 30 03:42:45 <NAiL> Welcome to the 266club, rwhitby ;-)
Jun 30 03:42:53 <rwhitby-away> dyoung: I'm not equipped to replace it on devslug at home
Jun 30 03:43:06 * rwhitby-away remove it with a 1cm tip
Jun 30 03:43:18 <dyoung> rwhitby-away, comfort.  
Jun 30 03:43:19 <NAiL> 1cm?
Jun 30 03:43:27 <dyoung> CM?!
Jun 30 03:43:28 <dyoung> holy cow
Jun 30 03:43:32 <dyoung> Brave man
Jun 30 03:43:38 <rwhitby-away> well, maybe 5mm
Jun 30 03:43:40 <NAiL> I thought my tip was large enough
Jun 30 03:43:44 <rwhitby-away> but big, dirty and ugly
Jun 30 03:43:49 <dyoung> Actually,  thats one of the reasons why I lifted one side only.
Jun 30 03:43:56 <dyoung> in case it didnt work. heh
Jun 30 03:44:02 <rwhitby-away> dyoung: do you want to know the even worse news?
Jun 30 03:44:03 <dyoung> oh ye of little faith.
Jun 30 03:44:12 <dyoung> Uh, sure.
Jun 30 03:44:17 <rwhitby-away> devslug was PinguSlug
Jun 30 03:44:24 <dyoung> Nooooooooooooooo!
Jun 30 03:44:41 <rwhitby-away> so now you have to help me get Samba working on OpenSlug :-)
Jun 30 03:45:00 <NAiL> huh? Samba works on openslug. ipkg install samba. :)
Jun 30 03:45:31 <NAiL> (well, as soon as I resubmit all the patches)
Jun 30 03:45:42 <dyoung> how long was that?  like 8 min or so?
Jun 30 03:45:54 <dyoung> Max Temp Case open was 47DegC
Jun 30 03:46:10 <dyoung> loadavg 2
Jun 30 03:46:10 <NAiL> dyoung: Did you measure on the back?
Jun 30 03:46:13 <dyoung> no
Jun 30 03:46:16 <dyoung> chekcing
Jun 30 03:46:16 <NAiL> try 
Jun 30 03:46:24 <rwhitby-away> ok, devslug is up now
Jun 30 03:46:38 <rwhitby-away> (devslug is the one with the 43pin, and is a bit temperamental at times)
Jun 30 03:46:41 <peteru-home> rwhitby-away: How did you resurect it?
Jun 30 03:46:46 <rwhitby-away> so who else has two SpeedySlugs ?
Jun 30 03:46:47 <dyoung> 42deg under the cpu
Jun 30 03:46:55 <NAiL> ok, not that bad
Jun 30 03:47:00 <NAiL> rwhitby-away: noone. Yet. 
Jun 30 03:47:02 <rwhitby-away> peteru-home: sometimes it just takes a few goes to make it boot
Jun 30 03:47:06 <dyoung> Bah
Jun 30 03:47:09 <rwhitby-away> NAiL: someone does now :-)
Jun 30 03:47:10 <NAiL> I wont get another until tomorrow :(
Jun 30 03:47:10 <dyoung> rwhitby is the frst with 2
Jun 30 03:47:22 <NAiL> rwhitby-away: So? I had one hours before you ;)
Jun 30 03:47:42 <NAiL> I've had time to compile perl three times over (almost) ;-)
Jun 30 03:47:46 <dyoung> I guess its pointless to SpeedySlug-ify Lennerts slug.
Jun 30 03:47:48 <rwhitby-away> NAiL: I'm just creating the database so that fact can be recorded for posterity ...
Jun 30 03:48:04 <NAiL> hehe
Jun 30 03:48:22 <rwhitby-away> what's the official generic name for SpeedySlugs ?  Is it SpeedySlug or something else?
Jun 30 03:48:26 <rwhitby-away> (for the database name)
Jun 30 03:48:39 <NAiL> I've gotta go to work. Remeber, #2 is mine ;)
Jun 30 03:48:43 <dyoung> TurbotSlug?
Jun 30 03:48:46 <NAiL> Remember, even
Jun 30 03:48:48 <dyoung> TurboSlug?
Jun 30 03:48:52 <dyoung> FastSlug?
Jun 30 03:48:55 <rwhitby-away> TurboSlug
Jun 30 03:49:01 <dyoung> SpeedySlug is the cutest.
Jun 30 03:49:16 <NAiL> NotSluggish?
Jun 30 03:49:17 <NAiL> :P
Jun 30 03:49:30 <dyoung> are we letting this one out on the ML?
Jun 30 03:49:45 <dyoung> if so, it'll be a lot of fun with the verbage.
Jun 30 03:49:53 <dyoung> Is your slug Sluggish?
Jun 30 03:50:02 <dyoung> Wish it was speedier?
Jun 30 03:50:03 <NAiL> dyoung: Did you read the wiki page? ;)
Jun 30 03:50:19 <NAiL> Tired of sluggish performance? Want to squeeze a bit more out of your slug? Feel that you're not getting the dish you paid for?
Jun 30 03:50:49 <rwhitby-away> TurboSlug is more authoritative
Jun 30 03:50:58 <rwhitby-away> http://groups.yahoo.com/group/nslu2-linux/database?method=reportRows&tbl=11
Jun 30 03:51:01 <dyoung> I agree.
Jun 30 03:51:11 <rwhitby-away> going offline to do #3.
Jun 30 03:51:14 <rwhitby-away> back in a bit
Jun 30 03:51:29 <dyoung> I need to find some of that heatsink epoxy stuff
Jun 30 03:51:34 <dyoung> so I can glue a heatsink on
Jun 30 03:51:57 <dyoung> max temp is still < 47deg
Jun 30 03:52:34 * NAiL goes to work. Figure out how to make it even speedier in the meantime :-P
Jun 30 03:53:59 <jacques> my perl 'make' time went from real    89m2.520s to real    61m26.581s
Jun 30 03:54:17 <dyoung> Lots of I/O I guess eh.
Jun 30 03:54:34 <rwhitby-away> ok, that's three
Jun 30 03:54:39 <dyoung> but a 30% reduction in build time is nothing to sneeze at
Jun 30 03:54:53 <rwhitby> so, which three numbers do I take in the database?
Jun 30 03:55:10 <dyoung> Bah, only one database entry per member. :-)
Jun 30 03:55:10 <jacques> going to time the tests now
Jun 30 03:55:19 <NAiL> rwhitby: 567
Jun 30 03:55:51 <NAiL> jacques: I guess most of the time gained is in configure. It doesn't swap much. During make it swaps a lot more
Jun 30 03:56:04 <rwhitby> dyoung: bah - one entry per TurboSlug
Jun 30 03:56:14 <dyoung> Grr
Jun 30 03:56:14 <jacques> NAiL, only so much time can be gained in configure since it's short
Jun 30 03:56:34 * dyoung start ripping open slugs
Jun 30 03:56:59 <jacques> NAiL, I suspect the 'class' test dominates everything in tests, and it swaps like mad
Jun 30 03:57:02 <dyoung> actually not.
Jun 30 03:57:02 <jacques> thrashes actually
Jun 30 03:57:08 <NAiL> It's now we find out that the modded slugs last for 6 hours before dying
Jun 30 03:57:18 <NAiL> ;)
Jun 30 03:57:19 <dyoung> I'm going to be slow on this one
Jun 30 03:57:37 <dyoung> I'll let devslug run this CPU intensive task for a while
Jun 30 03:57:47 <dyoung> and if it still works in the morning then I'll mod the other 2
Jun 30 03:57:55 <rwhitby> I'm gonna be the clueless user who just does the mods and forgets about it
Jun 30 03:57:57 <ep1220> just studying the datasheet, page 81; says max Casetemp can be 113C for a 266MHz device.
Jun 30 03:58:11 <dyoung> ep1220, Wow really?
Jun 30 03:58:13 <dyoung> thats great!
Jun 30 03:58:31 <dyoung> My CPU Intensive task has been going for 20min or so, the max temp is still < 47deg
Jun 30 03:58:33 <NAiL> Will the plastic handle that?
Jun 30 03:58:33 <rwhitby> dyoung: now to restart that pingu job on devslug ...
Jun 30 03:58:41 <NAiL> It's enough to cook water 
Jun 30 03:58:42 <ep1220> intel DOc 25247905
Jun 30 03:59:16 <dyoung> These only one spot on the chip thats  > 47dg
Jun 30 03:59:23 <dyoung> 5mm to the right of the i
Jun 30 03:59:35 <ep1220> they say important point is the center of the device
Jun 30 04:00:06 <jacques> mmmm ribeye
Jun 30 04:00:11 <rwhitby> dyoung: if the pingu dvd succeeds, then that's a good enough test for me :-)
Jun 30 04:01:41 <jacques> ep1220, is that for regular or milspec?
Jun 30 04:02:03 <ep1220> Commerical speced device 0..70C)
Jun 30 04:02:22 <ep1220> asumes 2W consumed by the CPU
Jun 30 04:07:12 <rwhitby> [g2]: get out your soldering iron ....
Jun 30 04:08:01 <[g2]> hey :)
Jun 30 04:08:17 <[g2]> ok
Jun 30 04:08:22 <rwhitby> you can get #8 is you're quick ...
Jun 30 04:08:39 <[g2]> :)
Jun 30 04:09:06 <[g2]> now it'll be 80-hour weeks
Jun 30 04:09:08 <[g2]> :)
Jun 30 04:09:16 <[g2]> slugtime * 2
Jun 30 04:09:50 <[g2]> anybody measure heat or power draw differences ?
Jun 30 04:10:29 <rwhitby> heh - dyoung must be upgrading his miau slug ...
Jun 30 04:11:04 <[g2]> the *real* question is who's gonna be the first to put a heat sink on there and see if it OC's to 400 :)
Jun 30 04:11:51 <rwhitby> [g2]:it wont - fuses on chip prevent that
Jun 30 04:12:07 <[g2]> you've tried ?
Jun 30 04:13:22 <jacques> NAiL, forget what I said about the class test - it takes a long time but doesn't thrash - I must be remembering a python test - yeah that's it
Jun 30 04:13:26 <[g2]> or I guess from the manual all the configs are the same
Jun 30 04:13:40 <[g2]> except for the 1/2 speed
Jun 30 04:14:14 <[g2]> jacques, how are the benchmarks running ?
Jun 30 04:14:36 <jacques> perl configure takes 1/2 the time, make takes 67% the time
Jun 30 04:14:49 <[g2]> cool
Jun 30 04:14:51 <jacques> running perl tests now
Jun 30 04:15:14 <[g2]> I'd think that some things like the make would be partially I/O bound
Jun 30 04:15:36 <jacques> yep
Jun 30 04:15:54 <[g2]> I guess that's where the anticipatory I/O would make a difference
Jun 30 04:16:32 <rwhitby> [g2]: the manual talks about fuses
Jun 30 04:16:45 <[g2]> cool!
Jun 30 04:17:45 <[g2]> so I guess the next thing is to OC the FSB :)
Jun 30 04:18:07 <[g2]> xtal or CLK_IN (or whatever it's really called )
Jun 30 04:18:22 <jacques> do we know the ram spec?
Jun 30 04:18:36 <[g2]> I think it's 133
Jun 30 04:18:48 <[g2]> I'm sure there so wiggle room
Jun 30 04:18:54 <[g2]> some
Jun 30 04:19:51 * [g2] wouldn't be surprised if the 420/422/425 are all the same die
Jun 30 04:35:21 <[g2]> great work all around!
Jun 30 04:35:45 <[g2]> NAiL, I guess you didn't beleive me about the alignment traps :)
Jun 30 04:36:53 <[g2]> on the perl configure
Jun 30 04:39:25 <dyoung> stupid cable modem went dead
Jun 30 04:39:49 <[g2]> doh! too many packets from the turbo slug
Jun 30 04:40:08 <dyoung> anyways through all of that, the maxtemp is still 47deg
Jun 30 04:40:42 <dyoung> anyone know if beewoolie has a yahoo id?
Jun 30 04:41:37 <dyoung> I added the record for him.
Jun 30 04:42:22 <[g2]> you/he can edit it later :)
Jun 30 04:42:41 <[g2]> so just remove R83 then eh ?
Jun 30 04:43:23 <dyoung> I can edit it later.
Jun 30 04:43:47 <dyoung> hehe I should put some comments in there
Jun 30 04:44:09 <dyoung> [g2] yeah, remove R83
Jun 30 04:44:51 <rwhitby> my three slugs don't seem to be running any hotter
Jun 30 04:45:36 <dyoung> Devslug is still at 47deg
Jun 30 04:46:21 <jacques> mine doesn't feel any hotter either
Jun 30 04:46:57 <jacques> OK, perl 'make test' went from ~ 75 minutes to real    50m17.533s
Jun 30 04:47:01 <jacques> I see a trend
Jun 30 04:47:09 <rwhitby> 2/3
Jun 30 04:47:12 <jacques> yp
Jun 30 04:47:22 <jacques> (that's 2/3 of yep)
Jun 30 04:47:55 <jacques> All tests successful.   I didn't get the failure NAiL did
Jun 30 05:00:18 <dyoung> devslug still at 47deg
Jun 30 05:01:16 <rwhitby> how hot can a 266MHz processor get anyway :-) it's a slug, don't forget.
Jun 30 05:01:44 <dyoung> its been going for a while, I'm pretty confident it wont get hotter than 47deg
Jun 30 05:01:53 <dyoung> in a case, dunno
Jun 30 05:03:00 <ep1220> dyoung: Cover it with a piece of paper and periodically check ..
Jun 30 05:03:56 <dyoung> I got a cd jewel case sitting on it now. :-)
Jun 30 05:06:03 <rwhitby> I've got all three back in the cases and up on the slug-central shelf.
Jun 30 05:06:18 <rwhitby> dyoung: is ciabot_monotone working?
Jun 30 05:07:16 <dyoung> no.  For unknown reasons when I run it by hand it fails with a monotone error
Jun 30 05:08:56 <rwhitby> almost finished setting up the new db ...
Jun 30 05:13:52 <dyoung> 48.5 deg with a CD Jewel case sitting on top of slug
Jun 30 05:15:10 <dyoung> AFAICT, thats around the same as the internal temps of a stock slug
Jun 30 05:30:06 <[g2]> Processor       : XScale-IXP42x Family rev 1 (v5b)
Jun 30 05:30:06 <[g2]> BogoMIPS        : 266.24
Jun 30 05:30:27 <rwhitby> [g2] grabs #8 ?
Jun 30 05:30:40 <[g2]> well the unit runs at 266
Jun 30 05:30:45 <[g2]> R83 gone
Jun 30 05:32:08 <[g2]> #8
Jun 30 05:32:27 <rwhitby> that completes the set of old-timers :-)
Jun 30 05:32:47 <[g2]> in less than 12 hours :)
Jun 30 05:32:56 <dyoung-zzzz> That special spot 5mm to the right of the "i" is 49.5deg now after sitting under a jewel case for a while.
Jun 30 05:32:57 <rwhitby> slugtime
Jun 30 05:33:36 <dyoung-zzzz> I'll let it run overnight, if it still works in the morning, I'm calling it stable
Jun 30 05:34:01 <rwhitby> ye of little faith
Jun 30 05:34:09 <[g2]> usually one can't put their finger on a chip
Jun 30 05:34:17 <dyoung-zzzz> my finger test passes.
Jun 30 05:34:19 <[g2]> I know the old moto's you couldn't
Jun 30 05:34:48 <dyoung-zzzz> Given the spec ep1220 mentioned earlier, its probably going to be just fine.
Jun 30 05:34:51 <[g2]> I've seen several ppl leave a little skin on chips
Jun 30 05:36:51 <ep1220> g2: Does the Avila use a heatsink ?
Jun 30 05:37:20 <[g2]> no :)
Jun 30 05:37:35 <[g2]> g'day ep1220 
Jun 30 05:37:49 <[g2]> are you speed enhanced ?
Jun 30 05:37:56 <ep1220> g2: not yet.
Jun 30 05:38:12 <ep1220> Ithink what i want to measure before and after ..
Jun 30 05:38:21 <[g2]> nod
Jun 30 05:38:41 <[g2]> I think many of us have a couple units
Jun 30 05:39:04 <ep1220> i own just 1
Jun 30 05:39:10 * rwhitby can't measure before, cause all 3 are TurboSlugs
Jun 30 05:39:26 <[g2]> rwhitby, always runs so fast :)
Jun 30 05:39:44 <rwhitby> heh, if one blows up, the community will replace it :-0
Jun 30 05:40:21 <rwhitby> if all three blow up, then we will know it's not safe, and I have saved lots of people heartache
Jun 30 05:40:38 <rwhitby> (and the community will buy three more)
Jun 30 05:43:05 <ep1220> there is another device to be watched:
Jun 30 05:43:24 <ep1220> the voltage regulator which genertaes the 1.3V core-voltage
Jun 30 05:43:40 <ep1220> that one could now have to spupply 100% more current
Jun 30 05:44:11 <[g2]> do we have a part number ?
Jun 30 05:44:20 <ep1220> me, not yet
Jun 30 05:45:27 <VoodooZ_Work> Hey guys
Jun 30 05:45:31 <VoodooZ_Work> ANy news?
Jun 30 05:45:35 <[g2]> howdy
Jun 30 05:45:46 <rwhitby> VoodooZ_Work: get yer soldering iron hot
Jun 30 05:45:47 <VoodooZ_Work> DO i smell burning slugs? :)
Jun 30 05:45:51 <VoodooZ_Work> NO way!!
Jun 30 05:45:56 <VoodooZ_Work> Does it work?
Jun 30 05:46:06 <[g2]> http://groups.yahoo.com/group/nslu2-linux/database?method=reportRows&tbl=11
Jun 30 05:46:07 <rwhitby> http://groups.yahoo.com/group/nslu2-linux/database?method=reportRows&tbl=11
Jun 30 05:46:10 <rwhitby> snap
Jun 30 05:46:12 <VoodooZ_Work> OH YEAH!!
Jun 30 05:46:24 <jacques> python make went from real    52m5.662s to real    35m57.018s
Jun 30 05:46:27 <VoodooZ_Work> damn, I missed all the excitment again!
Jun 30 05:46:34 <VoodooZ_Work> Yes sir!
Jun 30 05:46:36 <[g2]> LOL
Jun 30 05:46:37 <rwhitby> VoodooZ_Work: if you're quick, you can get #9
Jun 30 05:47:00 <VoodooZ_Work> hehehe. I doubt it as I just got in work and I'm leaving for camping right after. that sucks!
Jun 30 05:47:08 <dyoung-zzzz> VoodooZ_Work, Told you.  10hrs. :-)
Jun 30 05:47:17 <VoodooZ_Work> hehehe What is required?
Jun 30 05:47:22 <VoodooZ_Work> I'
Jun 30 05:47:28 <dyoung-zzzz> VoodooZ_Work, Remove R83
Jun 30 05:47:31 <VoodooZ_Work> I'm too excited to read right now. :)
Jun 30 05:47:36 <VoodooZ_Work> That's it?
Jun 30 05:47:38 <[g2]> simple even [g2] did it :)
Jun 30 05:47:43 <VoodooZ_Work> cool!
Jun 30 05:47:52 <VoodooZ_Work> What about power consumption/heat?
Jun 30 05:47:57 <rwhitby> bah, who cares
Jun 30 05:48:05 <VoodooZ_Work> I already have heatsinks installed. I knew they would come-in handy!
Jun 30 05:48:21 <dyoung-zzzz> I've been measureing mine for a few  hours, it maxes out at 49.5 no case; but with a CD Jewel case sitting on it
Jun 30 05:48:23 <VoodooZ_Work> I run on batteries so I might need to compare.
Jun 30 05:48:31 <jacques> hmm, I was feeling my case and it's soft now, and very hot - it's like sagging
Jun 30 05:48:37 <VoodooZ_Work> hehe
Jun 30 05:48:41 <jacques> j/k  :-D
Jun 30 05:48:44 <dyoung-zzzz> (thats 49.5degC btw
Jun 30 05:48:46 <dyoung-zzzz> )
Jun 30 05:48:51 <dyoung-zzzz> not current.
Jun 30 05:48:55 <VoodooZ_Work> I know.
Jun 30 05:49:15 <VoodooZ_Work> I guess my robot would be constantly running at 100% so it wouldn't be idling much.
Jun 30 05:49:30 <dyoung-zzzz> These tests are with the CPU at loadavg of 2+
Jun 30 05:49:37 <rwhitby> VoodooZ_Work: you could put a switch on it :-)
Jun 30 05:49:40 <VoodooZ_Work> Anybody had difficulties? As in problems with sub-par parts.
Jun 30 05:50:03 <VoodooZ_Work> yeah, I was thinking about that. I wonder if I can switch the speed while it's running?
Jun 30 05:50:21 <[g2]> don't think so
Jun 30 05:50:31 <VoodooZ_Work> probably not safe anyways.
Jun 30 05:50:34 <dyoung-zzzz> unless you twiddle the master clock.
Jun 30 05:50:39 <dyoung-zzzz> but its probably not a good idea.
Jun 30 05:50:40 <VoodooZ_Work> scary!
Jun 30 05:50:49 <[g2]> the values are latched at boot
Jun 30 05:50:55 <VoodooZ_Work> so what's the slug name for this monster we've created?
Jun 30 05:51:03 <VoodooZ_Work> [g2]: true.
Jun 30 05:51:47 <rwhitby> TurboSlug
Jun 30 05:51:47 <VoodooZ_Work> wow! that's fantastic! I'll be able to process probably twice the amount of video frames per second now!
Jun 30 05:51:51 <VoodooZ_Work> I like it!
Jun 30 05:52:13 <VoodooZ_Work> too funny, all 3 channels are talking about it!
Jun 30 05:52:21 <rwhitby> another dyoung name ...
Jun 30 05:52:25 <dyoung-zzzz> VoodooZ_Work, Dont talk about iti on nslu2-linux yet.
Jun 30 05:52:42 <VoodooZ_Work> hopefully people will update the performance wiki pages with their benchmarks.
Jun 30 05:52:49 <VoodooZ_Work> seriously?
Jun 30 05:53:13 <VoodooZ_Work> Nail already did./
Jun 30 05:53:19 <rwhitby> mailing lis
Jun 30 05:53:21 <rwhitby> list
Jun 30 05:53:38 <[g2]> I guess the cat's out of the bag :)
Jun 30 05:53:56 <dyoung-zzzz> I see, he did.
Jun 30 05:54:02 <dyoung-zzzz> oh well. so much with that plan. :-)
Jun 30 05:54:06 <VoodooZ_Work> hehehe. Hard not to be excited about that. I might pick up an extra one before they all disappear.
Jun 30 05:54:23 <rwhitby> only on the IRC channel.  less than 1% of the community.
Jun 30 05:55:04 <dyoung-zzzz> VoodooZ_Work, me too
Jun 30 05:55:15 <dyoung-zzzz> I have 3 in my house.
Jun 30 05:55:19 <rwhitby> but anyone checking RecentChanges on the wiki will find it
Jun 30 05:55:21 <dyoung-zzzz> need backup.
Jun 30 05:55:37 <dyoung-zzzz> and anyone that is chekcing recent changes probably deserves to find it.
Jun 30 05:55:54 <VoodooZ_Work> yep
Jun 30 05:56:00 <rwhitby> we will probably want to password protect that page
Jun 30 05:56:18 <rwhitby> don't want someone nefariously changing the resistor number ....
Jun 30 05:56:19 <VoodooZ_Work> I don't think Linksys would like that little tweak much. 
Jun 30 05:56:36 <VoodooZ_Work> we are so BAD!
Jun 30 05:56:40 <dyoung-zzzz> and someone should probablyl write a bit for "why this works".
Jun 30 05:56:42 * dyoung-zzzz hides
Jun 30 05:57:04 <VoodooZ_Work> And some benchmarks to prove it.
Jun 30 05:57:13 <[g2]> someone should email Linksys why it's set that way
Jun 30 05:57:22 <VoodooZ_Work> btw I come back from camping there won't be anything to benchmark though. dang!
Jun 30 05:57:36 <dyoung-zzzz> Linksys: "I dunno, it came from the factory like that"
Jun 30 05:57:47 <VoodooZ_Work> probably because the box worked perfectly fine at 133Mhz.
Jun 30 05:58:20 <[g2]> I'd guess they didn't want to compete with other products from Cisco/Linksys
Jun 30 05:58:23 <VoodooZ_Work> Or maybe their power design fell short.
Jun 30 05:58:31 <[g2]> or the "turbo" version could come out
Jun 30 05:58:32 <VoodooZ_Work> makes sense.
Jun 30 05:58:57 * VoodooZ_Work raises his right hand for a high-five.
Jun 30 05:59:16 <dyoung-zzzz> well if noone appends a "why does this work" section in the next 8hrs to that page, I'll write something.
Jun 30 05:59:20 <dyoung-zzzz> I need to sleep.
Jun 30 05:59:28 <VoodooZ_Work> goodnight. 
Jun 30 05:59:37 <[g2]> sweet dreams speedy
Jun 30 05:59:37 <rwhitby> I added the PayPal buttons :-)
Jun 30 05:59:52 <[g2]> NAiL, ping
Jun 30 06:00:07 <dyoung-zzzz> Classic.
Jun 30 06:01:10 <dyoung-zzzz> [g2] nail mumbled something about going to work after staying up all night fooling with TurboSlug.
Jun 30 06:01:16 <rwhitby> ep1220, bewoolie or NAiL should be the one to announce it.  probably NAiL cause he was the first to do the mod.  Or Jacques, cause he was the first to do the official mod :-)
Jun 30 06:01:33 <[g2]> nod
Jun 30 06:01:42 <dyoung-zzzz> hehe
Jun 30 06:01:51 <dyoung-zzzz> I always get thwarted at the last minute.
Jun 30 06:02:00 <jacques> as long as I get my prize I don't care who announces it
Jun 30 06:02:01 <dyoung-zzzz> 2.6.11, turboslug, ....
Jun 30 06:02:13 <VoodooZ_Work> what was the magic bit combination after all? I guess my lack logic worked.
Jun 30 06:02:20 <rwhitby> 011
Jun 30 06:02:23 <VoodooZ_Work> cool.
Jun 30 06:02:31 <[g2]> dyoung could also announce it
Jun 30 06:02:36 <VoodooZ_Work> I knew the table in the DS was "lying" a bit.
Jun 30 06:02:37 <dyoung-zzzz> No
Jun 30 06:02:43 <dyoung-zzzz> I dont need to announce it
Jun 30 06:02:48 <rwhitby> VoodooZ_Work: 011 is documented in the table
Jun 30 06:02:53 <dyoung-zzzz> I already cast my vote for annoucer.
Jun 30 06:02:55 <VoodooZ_Work> How about you make them pay to get it? :)
Jun 30 06:03:05 <VoodooZ_Work> it is? sorry I thought it was the other ones.
Jun 30 06:03:07 <dyoung-zzzz> as long as I get my prize too. :-)
Jun 30 06:03:27 * VoodooZ_Work hugs dyoung-zzzz, ep1220, NAiL and [g2]
Jun 30 06:03:28 <dyoung-zzzz> and snowshoes.
Jun 30 06:03:42 <dyoung-zzzz> VoodooZ_Work, dont forget beewoolie
Jun 30 06:03:43 * VoodooZ_Work and whoever else participated
Jun 30 06:03:44 * [g2] cheers up
Jun 30 06:03:46 <VoodooZ_Work> sorry
Jun 30 06:03:57 <[g2]> at least VoodooZ_Work still cares about me
Jun 30 06:04:09 <VoodooZ_Work> you guys made me (a cheap bastard who wants more for less) a real happy fellow!
Jun 30 06:04:15 <VoodooZ_Work> Of course!
Jun 30 06:04:16 <VoodooZ_Work> :)
Jun 30 06:04:25 <[g2]> NAiL, doesn't talk to me anymore, leaves me off the cred's list.....
Jun 30 06:04:49 <rwhitby> [g2]: those like you who did all the work leading up to it are always forgotten :-)
Jun 30 06:04:52 <VoodooZ_Work> I'm now considering going home for lunch to remove a resistor! :)
Jun 30 06:05:00 <rwhitby> you gotta be around to get the glory :-)
Jun 30 06:05:28 <VoodooZ_Work> Yeah, that sucks I couldn't have been there as I would have sacrificed my slug anytime for the team.
Jun 30 06:05:35 <[g2]> i verify the actually pull configs from C4000020 and it's all forgotton
Jun 30 06:06:36 <VoodooZ_Work> so there's two ways to o/c that mofo? R64 and R83?
Jun 30 06:06:37 <rwhitby> I vote [g2] announces :-)
Jun 30 06:06:54 <rwhitby> VoodooZ_Work: yes, but only R83 is a documented option
Jun 30 06:06:57 <VoodooZ_Work> It would make a good addition to the newsletter! In big bold letters
Jun 30 06:07:01 <VoodooZ_Work> cool.
Jun 30 06:07:10 <rwhitby> [g2]: how about a special edition of SlugNews !!!!!
Jun 30 06:07:11 <[g2]> LOL
Jun 30 06:07:28 <VoodooZ_Work> If nobody is done by the time I come back from camping I'll add a picture of the resitor in question to the wiki
Jun 30 06:07:41 <rwhitby> yeah, you gotta do it.  SlugNews issue #2.  TurboSlugs arrive!
Jun 30 06:07:51 <dyoung-zzzz> wiki fixed
Jun 30 06:08:22 <[g2]> dyoung-zzzz, I'm just teasing NAiL :)
Jun 30 06:08:27 <VoodooZ_Work> I see Nail created a wiki page for the procedure already?
Jun 30 06:08:39 <[g2]> I wasn't serious
Jun 30 06:08:42 <rwhitby> [g2]: I'm serious about SlugNews being the announcement vehicle
Jun 30 06:08:57 <jacques> python tests are idiotic - the process gets bigger and bigger as it goes along
Jun 30 06:09:06 <jacques> til it's like > 60MB
Jun 30 06:09:11 <VoodooZ_Work> I just hope I get a fluke board which can't run at 266. that would suck.
Jun 30 06:09:18 <[g2]> AHH....
Jun 30 06:09:20 <rwhitby> Who's gonna have the first TurboFatSlug ?
Jun 30 06:09:32 <dyoung-zzzz> [g2] hands down on the TurboFatSlug
Jun 30 06:09:37 <[g2]> now you know why you don't see the 2x
Jun 30 06:09:40 <dyoung-zzzz> because hes got th eonly fatslug
Jun 30 06:09:43 <[g2]> you're swapping
Jun 30 06:10:01 <jacques> 88.1% wa - dubmasses
Jun 30 06:10:23 <rwhitby> [g2]: make your FatSlug into a TurboFatSlug, and do a SlugNews which announces FatSlugs, TurboSlugs, Apex, and TurboFatSlugs.
Jun 30 06:10:35 <jacques> huh?
Jun 30 06:10:38 <VoodooZ_Work> anybody measured openslug boot time?
Jun 30 06:10:46 <jacques> what did you mean by <[g2]> now you know why you don't see the 2x
Jun 30 06:10:54 <[g2]> rwhitby, my Avila board is a fatslug
Jun 30 06:11:04 <rwhitby> [g2]: doesn't count - it's not a slug
Jun 30 06:11:21 <VoodooZ_Work> yeah, I'm so excited I might just fatslug and apex mine!
Jun 30 06:11:25 <jacques> I wouldn't expect 2x performance from just doubling the core speed, except for processes which fit in cache
Jun 30 06:11:27 <rwhitby> it's a FatAvila
Jun 30 06:11:57 <jacques> only the python tests are serious swapping, everything else so far has been fine
Jun 30 06:11:58 <dyoung-zzzz> jacques, nod.  Plus the PCI is the same speed so all of the USB transactions are the same speed (more or less)
Jun 30 06:12:25 <jacques> yeah RAM and disk speed are same as before
Jun 30 06:12:28 <dyoung-zzzz> I'm pretty sure the network interface should be faster.
Jun 30 06:12:40 <jacques> hmmm how to test that properly...
Jun 30 06:12:44 <[g2]> nod RAM and USB would still be at 33Mhz
Jun 30 06:12:51 <[g2]> well RAM at 133
Jun 30 06:13:25 <dyoung-zzzz> I suppose one could rsync a gimangous file.
Jun 30 06:13:31 <jacques> this mod dues appear to make dropbear more usable speed-wise tho
Jun 30 06:13:40 <jacques> I'm sure I'll still get the disconnects
Jun 30 06:14:12 <dyoung-zzzz> ah well, I'll check back in a few hours.  Must sleep.
Jun 30 06:14:23 <jacques> later dyoung-zzzz 
Jun 30 06:14:30 <rwhitby> night dyoung-zzzz
Jun 30 06:15:35 <VoodooZ_Work> I'll post computer vision benchmarks later. this pretty much stresses the whole slug. USB2.0 high-speed mode, RAM and cpu.
Jun 30 06:16:36 <dyoung-zzzz> hmm,
Jun 30 06:16:43 <dyoung-zzzz> thats an interesting point.
Jun 30 06:16:44 <jacques> there's so much thrashing in the python tests that I think they'll blow the curve
Jun 30 06:17:20 <dyoung-zzzz> will the faster core speed allow these spurious EHCI interrupts to be handled?
Jun 30 06:17:26 <jacques> nope
Jun 30 06:17:31 <dyoung-zzzz> Bummers.
Jun 30 06:17:33 <jacques> I'm getting a ton of them right now
Jun 30 06:17:42 <VoodooZ_Work> This is killing me. I want to play with TurboSlug too! :)
Jun 30 06:18:01 <rwhitby> VoodooZ_Work: quit your job.  you know you want to ...
Jun 30 06:18:12 <dyoung-zzzz> VoodooZ_Work, dont worry, I missed almost all of it too.
Jun 30 06:18:16 <VoodooZ_Work> heheheh. I've been wanting to do that!
Jun 30 06:18:27 <dyoung-zzzz> I was busy getting my ass kicked.
Jun 30 06:18:31 <jacques> I feel like I'm abusing the poor slug whenever I do this
Jun 30 06:18:35 <VoodooZ_Work> Yeah, but now I want to benchmark my vision code with it.
Jun 30 06:18:42 <VoodooZ-boss> VoodooZ-boss: VoodooZ_Work: get back to work
Jun 30 06:18:43 <VoodooZ_Work> abuse away!
Jun 30 06:18:48 <VoodooZ_Work> hehehe
Jun 30 06:18:58 <jacques> well, abusing my poor hard drive
Jun 30 06:19:04 <VoodooZ_Work> :(
Jun 30 06:19:14 <VoodooZ_Work> jacques: what's that? you are getting more IRQ26 errors?
Jun 30 06:19:35 <jacques> more than what?
Jun 30 06:20:00 <jacques> more than before? no
Jun 30 06:20:14 <VoodooZ_Work> ok.
Jun 30 06:20:37 <rwhitby> he's just getting them faster now :-)
Jun 30 06:20:38 <VoodooZ_Work> I'm wondering as I'm using a USB2.0 camera in High speed mode so I'm hoping there won't be any side-effects.
Jun 30 06:20:41 <VoodooZ_Work> hehe
Jun 30 06:20:54 <VoodooZ_Work> I like the addition of the paypal button on the wiki page! too funny.
Jun 30 06:21:03 <dyoung-zzzz> thats just classic.
Jun 30 06:21:03 <VoodooZ_Work> I'll definitely use it.
Jun 30 06:21:20 <dyoung-zzzz> I'm not donating for the overclock mod.
Jun 30 06:21:31 <jacques> it's pretty much solid > 800% diskwait right now
Jun 30 06:21:38 <jacques> > 80%
Jun 30 06:22:08 <VoodooZ_Work> Ok, donation made.
Jun 30 06:22:09 <rwhitby> dyoung-zzzz: huh?
Jun 30 06:22:32 <dyoung-zzzz> I'm not going to paypal to the project for learning about the overclocking mod.
Jun 30 06:22:47 <VoodooZ_Work> but you would for the project as a whole right?
Jun 30 06:23:03 <dyoung-zzzz> I did.
Jun 30 06:23:06 <rwhitby> dyoung-zzzz: are you talking as dyoung or as joe user?
Jun 30 06:23:18 <dyoung-zzzz> As dyoung-zzzz of course.
Jun 30 06:23:27 <rwhitby> oh, no developer should ever donate.
Jun 30 06:23:31 <jacques> I feel same way
Jun 30 06:23:36 * VoodooZ_Work is confused.
Jun 30 06:23:57 <VoodooZ_Work> anwyays, I made a donation. irregardless.
Jun 30 06:24:01 <rwhitby> anyone who contributes already in some significant way, should not feel at all compelled to donate.
Jun 30 06:24:09 <rwhitby> but we won't stop them :-
Jun 30 06:24:11 <rwhitby> :-)
Jun 30 06:24:49 <rwhitby> and dyoung-zzzz's contributions have been some of the most significant in the history of this project.
Jun 30 06:25:02 <VoodooZ_Work> agreed.
Jun 30 06:25:03 <dyoung-zzzz> Bah.
Jun 30 06:25:08 <dyoung-zzzz> I know nothing.
Jun 30 06:25:21 <dyoung-zzzz> but I am going to sleep.
Jun 30 06:25:22 <dyoung-zzzz> night
Jun 30 06:25:23 <rwhitby> dyoung-zzzz: who flashed the first binary-sed'd image?
Jun 30 06:25:24 <VoodooZ_Work> ~praise dyoung-zzzz
Jun 30 06:25:26 <purl> All hail dyoung-zzzz!
Jun 30 06:25:45 <VoodooZ_Work> he certainly has guts!
Jun 30 06:25:57 <rwhitby> who always stayed up five hours later than me during peak development times
Jun 30 06:27:02 <rwhitby> wonder what ever happened to joule360 ...
Jun 30 06:27:16 <rwhitby> and mrthreeplates
Jun 30 06:27:41 <rwhitby> we need to let buzbee know about this mod ...
Jun 30 06:29:47 <rwhitby> ok, I sent Jim a note.
Jun 30 06:38:42 <ep1220> I tried to fix the link to the power IC on http://www.nslu2-linux.org/wiki/Info/LinksToHardwareComponentDocumentation
Jun 30 06:38:55 <ep1220> Now wiki wants a passwort to approve the URL (?)
Jun 30 06:39:15 <NAiL> [g2]: pong
Jun 30 06:39:21 <rwhitby> ep1220: done
Jun 30 06:39:22 <NAiL> dropping by for 5 mins
Jun 30 06:39:43 <ep1220> rwhitby: Thx.
Jun 30 06:39:49 <rwhitby> NAiL: do you want to do the announcement?
Jun 30 06:40:04 <NAiL> huh? On the ml? 
Jun 30 06:40:10 <rwhitby> yep
Jun 30 06:40:33 <NAiL> I'm probably going to be busy for most of the evening, so then it'll take a while
Jun 30 06:40:45 <NAiL> And, I'm not that much of a writer ;-)
Jun 30 06:41:18 <[g2]> NAiL, I was gonna tease you for leaving me off the credits :)
Jun 30 06:41:21 <rwhitby> you can defer to dyoung or [g2] (the SlugNews editor) if you like ... or jacques (he did the first official mod)
Jun 30 06:41:23 <rwhitby> your choice
Jun 30 06:42:24 <NAiL> [g2]: Wanna do it, since I managed to forget you?
Jun 30 06:42:39 <[g2]> no I was just teasing
Jun 30 06:42:58 <[g2]> I think you or beewoolie or ep1220 should do it
Jun 30 06:43:00 <VoodooZ_Work> don't worry [g2] I'm not on the credit either.
Jun 30 06:43:20 <[g2]> VoodooZ_Work, dyoung added me in 
Jun 30 06:43:21 <NAiL> Neither am I, even though I had the second one ;-)
Jun 30 06:43:29 <VoodooZ_Work> I did feed evil subliminal messages to NAiL though.
Jun 30 06:43:53 <VoodooZ_Work> oh well, I'll survive the ordeal.
Jun 30 06:44:21 <rwhitby> hey, it's a wiki - just add yourself in 
Jun 30 06:44:44 <[g2]> hey I *wanted* to tease NAiL about it
Jun 30 06:44:46 <VoodooZ_Work> hehehe.
Jun 30 06:44:49 <NAiL> hahah
Jun 30 06:45:07 <NAiL> I'm too much in a hurry to be teasable right now :p
Jun 30 06:45:13 <NAiL> (changing clothes while typing)
Jun 30 06:45:21 <[g2]> well CONGRATS! 
Jun 30 06:45:22 <rwhitby> too much information
Jun 30 06:45:29 <[g2]> have a great day
Jun 30 06:45:30 <VoodooZ_Work> :
Jun 30 06:45:37 <VoodooZ_Work> later
Jun 30 06:46:00 <rwhitby> NAiL: so who's gonna do the announcement?
Jun 30 06:46:03 <[g2]> where's the web cam when you need it
Jun 30 06:47:03 <NAiL> [g2]: Isn't it about time with a new slugnews edition? I think you should do it :D
Jun 30 06:47:48 <NAiL> are dyoung or jacques here?
Jun 30 06:48:00 <NAiL> ah, dyoung-zzz went to bed
Jun 30 06:48:17 <NAiL> bewoolie probably isn't up yet
Jun 30 06:48:38 <jacques> im here NAiL 
Jun 30 06:49:17 <[g2]> jacques, is I/O waiting :)
Jun 30 06:49:26 <rwhitby> Go jacques, go jacques, go jacques ...
Jun 30 06:49:51 <[g2]> Hurry up, wait.... Hurry up, wait ... Hurry up, wait ...
Jun 30 06:50:04 <jacques> it's getting to the thrash part of the second (cimpiled) phase of the python tests
Jun 30 06:50:13 <NAiL> heh, Jacques. You are the lucky one. Since you flashed the first official mod, you'll have the honor of doing the announcement. (Instead of the boots)
Jun 30 06:50:32 * NAiL gotta run, already late
Jun 30 06:50:34 <jacques> hmm, I'd really like to keep my perfect record of not posting to the list :-)
Jun 30 06:50:47 <rwhitby> jacques: it would make a good first post ...
Jun 30 06:50:54 <NAiL> Yeah :)
Jun 30 06:51:02 <rwhitby> I vote for jacques.
Jun 30 06:51:12 <rwhitby> then he can get his prize
Jun 30 06:51:21 <NAiL> He's got my vote
Jun 30 06:52:03 <[g2]> jacques, you've never posted to the list ?
Jun 30 06:52:12 <jacques> fine I'll do it if you guys help me
Jun 30 06:52:32 <rwhitby> [g2]: he never has
Jun 30 06:52:35 <jacques> I wanna make sure the post has the proper info and tone
Jun 30 06:52:49 <jacques> with the proper caveats :-)
Jun 30 06:52:58 <NAiL> Go on guys, help him a bit. I'll be with you in spirit (While eating at a resturant and shopping clothes)
Jun 30 06:53:17 <NAiL> Part of the info is in the howto already
Jun 30 06:53:29 * NAiL leaves
Jun 30 06:53:33 <NAiL> Thanks, jacques :)
Jun 30 06:53:42 <rwhitby> jacques: use the disclaimer at the bottom of this page: http://www.rwhitby.net/nslu2/serial.html
Jun 30 06:54:07 <jacques> rwhitby, cool
Jun 30 06:54:28 <rwhitby> oh. and you put it at the top and in the middle too :-)
Jun 30 06:54:41 <jacques> lol yeah!
Jun 30 06:55:17 <jacques> should I put a link to the intel appnote ?
Jun 30 06:55:58 <VoodooZ_Work> I'm trying to add a photo of the R83 to the wiki. How do I attach a photo?
Jun 30 06:56:34 <jacques> that function might require a password
Jun 30 06:56:46 <VoodooZ_Work> darn
Jun 30 06:57:20 <VoodooZ_Work> some other page used Attach:filename but how do I put the file on the page?
Jun 30 06:57:28 <jacques> I don't have it AFAIK or I would tell you
Jun 30 06:57:43 <VoodooZ_Work> np
Jun 30 06:58:05 <VoodooZ_Work> I'd like to put a photo to prevent people from removing the wrong one.
Jun 30 06:58:32 <VoodooZ_Work> rwhitby: what do you suggest I do?
Jun 30 06:58:57 <jacques> anyone object to my making the post a "special notice" ?
Jun 30 06:59:16 <rwhitby> jacques: that's appropriate
Jun 30 07:00:10 <rwhitby> VoodooZ_Work: send me the image and I'll put it in the uploads dir
Jun 30 07:00:26 <VoodooZ_Work> ok. I'll see if I still have your email around.
Jun 30 07:01:16 <jacques> cute subject or serious?  right now I have "TurboSlug on the loose"
Jun 30 07:04:27 <rwhitby> jacques: make sure people realise that it's only a 2x in CPU speed, and not on external interfaces, so they shouldn't expect too much
Jun 30 07:05:20 <[g2]> jacques, that should be "TurboSlugs
Jun 30 07:05:23 <peteru-home> btw: My understanding is that the NPE cores always run internally at 133MHz
Jun 30 07:05:31 <rwhitby> How about "TurboSlugs: Adventures in overclocking ..."
Jun 30 07:05:57 * peteru-home also thinks that he was the first person to use the name turboslug
Jun 30 07:06:20 <jacques> [g2], you're right - changed
Jun 30 07:06:35 <jacques> rwhitby, yes I will include that, along with some benchmarks
Jun 30 07:06:40 <rwhitby> peteru-home: did you coin that before today?
Jun 30 07:07:14 <peteru-home> Yes, when we were discussing overclocking via software a few months ago.
Jun 30 07:07:20 <jacques> ok subject changed to rwhitby's
Jun 30 07:07:23 <peteru-home> and again today, in #puppy
Jun 30 07:08:24 <rwhitby> indeed you did.  you beat dyoung by 30 minutes today :-)
Jun 30 07:08:36 <peteru-home> I think jbowler started this off when he observed that even though this is a 266MHz part, the timing suggested 133MHz operation.
Jun 30 07:09:52 <jacques> one of the first things we noticed when we booted was the 133MHz bogomips number
Jun 30 07:09:56 <peteru-home> at which point there was a great air of suspicion about Linksys engineering "crippling" our favourite toy
Jun 30 07:11:16 <peteru-home> anyway, thumbs up to you guys for following through and actually finding out how to "Unlock the power within."
Jun 30 07:14:34 <rwhitby> VoodooZ_Work: wiki updated with image.  nice job
Jun 30 07:15:26 <VoodooZ_Work> Thanks.
Jun 30 07:16:09 <rwhitby> check the bottom of the page too :-)
Jun 30 07:16:29 <VoodooZ_Work> What?
Jun 30 07:16:43 <VoodooZ_Work> My name is not there? 
Jun 30 07:16:58 <jacques> should I mention the specific resistor in the posting or just point ppl to the wiki page?
Jun 30 07:17:41 <rwhitby> just point to the wiki page, then if it does turn out to be a complete ballsup, people won't get the wrong info from the post
Jun 30 07:18:10 <jacques> cool
Jun 30 07:19:26 <ByronT-Away> FrankenSlug is headed to work with me - where the soldering iron is...
Jun 30 07:21:13 <jacques> not sure how to handle the attribution/thanking - right now I have "Pretty much the entire dev team contributed to this discovery."
Jun 30 07:21:50 <rwhitby> jacques: perfectly put
Jun 30 07:22:26 <rwhitby> Jim Buzbee got my email.
Jun 30 07:23:05 <rwhitby> ByronT is gunning for #9 ...
Jun 30 07:23:19 <VoodooZ_Work> I might be able to make 9-10 during my lunch.
Jun 30 07:23:19 <rwhitby> it'll be a race between ByronT and VoodooZ ...
Jun 30 07:23:24 <VoodooZ_Work> hehe
Jun 30 07:23:26 <VoodooZ_Work> damn right!
Jun 30 07:23:37 <rwhitby> 10 is so double figures ...
Jun 30 07:23:57 <VoodooZ_Work> hehe
Jun 30 07:24:07 <VoodooZ_Work> Maybe it will be an early lunch :)
Jun 30 07:24:27 <rwhitby> VoodooZ throws down the gauntlet ... what will ByronT do ... ?
Jun 30 07:25:29 <rwhitby> Remember you gotta see the 266 BogoMips in the dmesg before you can claim the number ...
Jun 30 07:25:49 <VoodooZ_Work> don't worry. I'll skip lunch to make sure it does! 
Jun 30 07:25:50 <ByronT-Away> actually FrankenSlug and ProdSlug are taking a rid
Jun 30 07:25:54 <ByronT-Away> +e
Jun 30 07:26:07 <VoodooZ_Work> I'll have benchmarks ready by the time I'm back.
Jun 30 07:26:08 <ByronT-Away> taking ProdSlug offline now
Jun 30 07:26:14 <rwhitby> wooo - ByronT is gonna push VoodooZ out of the top ten!
Jun 30 07:26:20 <VoodooZ_Work> hehehe
Jun 30 07:26:56 <VoodooZ_Work> oh well, better late than never. :)
Jun 30 07:27:12 <rwhitby> there's still time ...
Jun 30 07:27:15 <ByronT-Away> ProdSlug offline... time for shower and then to work!
Jun 30 07:27:17 <ByronT-Away> :)
Jun 30 07:27:29 <VoodooZ_Work> It's only 10:30 here so lunch is still far.
Jun 30 07:51:44 <rwhitby> ByronT, VoodooZ: Tiersten is going to beat you both ...
Jun 30 07:53:44 <VoodooZ_Work> damn, more competition! :)
Jun 30 07:54:03 <VoodooZ_Work> maybe 10 isn't that bad after all. 11 is way uncool though.
Jun 30 07:59:03 <ByronT-Away> <- heading into work
Jun 30 07:59:52 <jacques> I just added Dhrystone numbers to the howto page
Jun 30 07:59:59 <Tiersten> I would have been earlier but I've been away for the last day :|
Jun 30 08:00:47 <ByronT-Away> k, I am packed up and heading into work with two slugs to be turbo'd
Jun 30 08:01:06 <VoodooZ_Work> how far is work? :)
Jun 30 08:01:14 <ByronT-Away> 10 minutes
Jun 30 08:01:28 <VoodooZ_Work> dope! :)
Jun 30 08:02:08 <Tiersten> #9 added
Jun 30 08:02:43 <jacques> we already got a response to the list post
Jun 30 08:02:54 <Tiersten> Yeah
Jun 30 08:03:07 <Tiersten> You do wonder why Linksys/Sercomm decided to underclock it tho
Jun 30 08:03:24 <jacques> sure I do
Jun 30 08:05:08 <Tiersten> Maybe it's all part of their evil scheme.  Release a 133MHz one and then release a new improved 266MHz one :)
Jun 30 08:07:02 <peteru-home> With twice as much RAM, to really put the boot in.
Jun 30 08:07:31 <Tiersten> Yeah
Jun 30 08:07:49 <rwhitby> ~nefarious
Jun 30 08:07:50 <purl> i guess nefarious is to be infamous by way of nastyness
Jun 30 08:07:52 <Tiersten> The PCB does kind of look vaguely multipurpose.  The alternate positions for the LEDs for one
Jun 30 08:11:01 <rwhitby> night turbosluggers
Jun 30 08:11:11 <[g2]> #10
Jun 30 08:11:21 <VoodooZ_Work> night
Jun 30 08:11:25 <VoodooZ_Work> #10? who?
Jun 30 08:11:50 <[g2]> TurboFatSug
Jun 30 08:11:52 <[g2]> TurboFatSlug
Jun 30 08:11:54 <[g2]> me
Jun 30 08:12:04 <VoodooZ_Work> oh, I thought you were already in there,
Jun 30 08:12:14 <[g2]> that was a non fat slug
Jun 30 08:12:18 <VoodooZ_Work> i see.
Jun 30 08:12:27 <VoodooZ_Work> Thanks, now I'll be 12! :)
Jun 30 08:12:44 <Tiersten> #10 being a turbofatslug seems quite fitting
Jun 30 08:12:51 <[g2]> I want to see Perl run 2x
Jun 30 08:13:12 <VoodooZ_Work> yeah, I'd love to have a turbofatslug too.
Jun 30 08:13:22 <Tiersten> Need to get ka6sox to do this ASAP when he gets back to the native comp slug
Jun 30 08:14:12 <jacques> yeah both of them
Jun 30 08:14:30 <VoodooZ_Work> I can't wait to see if my image processing function will benefit from the speed boost. there's lots of memory access so it's not going to be 2x.
Jun 30 08:14:52 <[g2]> VoodooZ_Work, I'm building boards
Jun 30 08:15:05 <[g2]> I think you'll really like the board I building
Jun 30 08:17:06 <[g2]> hey this should cut my kernel compile time in half
Jun 30 08:17:48 <VoodooZ_Work> building boards? 
Jun 30 08:18:02 <VoodooZ_Work> for selling you mean or for work?
Jun 30 08:18:14 <[g2]> I'm having custom boards made
Jun 30 08:18:22 <[g2]> first run is 200
Jun 30 08:18:37 <VoodooZ_Work> what's on them?
Jun 30 08:18:49 <[g2]> 422 @ 266, 64MB
Jun 30 08:18:55 <VoodooZ_Work> general purpose dev boards? nice.
Jun 30 08:18:57 <[g2]> serial, jtag,
Jun 30 08:19:10 <[g2]> CF :)
Jun 30 08:19:14 <VoodooZ_Work> How much flash?
Jun 30 08:19:21 <[g2]> .5MB
Jun 30 08:19:30 <[g2]> just enough to boot of the CF
Jun 30 08:19:38 <VoodooZ_Work> it's a pity it's not an xscale with hw I2C.
Jun 30 08:19:52 <VoodooZ_Work> USB?
Jun 30 08:20:02 <[g2]> I2C should come out to a header
Jun 30 08:20:13 <VoodooZ_Work> software I2C you mean?
Jun 30 08:20:14 <[g2]> yes USB
Jun 30 08:20:17 <[g2]> nod
Jun 30 08:20:25 <VoodooZ_Work> How much do you sell those?
Jun 30 08:21:15 <[g2]> in a case with compliance it'll probably be near $249
Jun 30 08:21:33 <[g2]> there are some other goodies too
Jun 30 08:21:52 <VoodooZ_Work> The slug is cheaper for hobbyist like me
Jun 30 08:21:59 <[g2]> nod
Jun 30 08:22:07 <VoodooZ_Work> but I might buy a real good one for my next robot project. 
Jun 30 08:22:16 <[g2]> nod everyone wants to get out the soldering iron
Jun 30 08:22:21 <VoodooZ_Work> I'll need more juice and I2C/Can in h/w.
Jun 30 08:22:25 <[g2]> not
Jun 30 08:22:35 <VoodooZ_Work> true.
Jun 30 08:23:39 <VoodooZ_Work> or even an FPGA/ASIC that would preprocess the video stream. (stereo?!)
Jun 30 08:24:18 <VoodooZ_Work> ByronT-Away: So? Are you #11 yet?
Jun 30 08:24:20 <[g2]> they will probably be some room for expansion
Jun 30 08:24:43 <ByronT-Away> heh, you beat me on my way into work?
Jun 30 08:24:54 <VoodooZ_Work> [g2] did, not me.
Jun 30 08:25:06 <VoodooZ_Work> He's showing off with his TurboFatSlug! :)
Jun 30 08:25:22 <ByronT> ah...
Jun 30 08:25:40 <ByronT> well the soldering iron was plugged in before I unloaded my briefcase
Jun 30 08:25:46 <jacques> lol
Jun 30 08:25:48 <VoodooZ_Work> heheh
Jun 30 08:27:46 <Tiersten> Shouldn't take very long for #11 then :)
Jun 30 08:28:27 <VoodooZ_Work> ok, Maybe I should reserve #12. #13 is just not lucky.
Jun 30 08:28:31 <[g2]> jacques, how long did your perl config take ?
Jun 30 08:28:43 <jacques> i posted it here... looking
Jun 30 08:28:54 <ByronT> here goes nothing
Jun 30 08:29:24 <[g2]> I think there may be a minor problem with the time
Jun 30 08:29:29 <VoodooZ_Work> It's actually a good thing that I'm last to do it: I have lots of beta-testers!
Jun 30 08:29:44 <jacques> <jacques> previous value: real    18m53.431s
Jun 30 08:29:44 <jacques> <jacques> now: real    10m7.199s
Jun 30 08:29:50 <Tiersten> time?
Jun 30 08:30:10 <[g2]> date
Jun 30 08:30:17 <Tiersten> Like how?
Jun 30 08:30:20 <[g2]> the time on the device
Jun 30 08:30:30 <VoodooZ_Work> because of the overclock you mean?
Jun 30 08:30:31 <[g2]>  date
Jun 30 08:30:31 <[g2]> Tue Nov 30 01:57:02 UTC 1999
Jun 30 08:30:44 <VoodooZ_Work> I hope it doesn't mess up the I2C timing.
Jun 30 08:30:51 <Tiersten> Whats wrong with that?
Jun 30 08:31:11 <Tiersten> Mine booted up fine and has the correct date & time.  It's a couple of minutes off but it's always like that
Jun 30 08:31:19 * [g2] I'm gonna party like it's 1999
Jun 30 08:31:20 <VoodooZ_Work> good
Jun 30 08:31:29 <Tiersten> Thu Jun 30 15:27:29 IST 2005
Jun 30 08:31:42 <Tiersten> No clue what it thinks it is IST but the date and time are roughly correct
Jun 30 08:31:47 <Tiersten> why it thinks
Jun 30 08:34:02 <Tiersten> IST = Irish Summer Time apparently so it's correct from the point of view about the time since IST = +1 like BST which it should be
Jun 30 08:34:12 <[g2]> real    8m 13.27s
Jun 30 08:34:12 <[g2]> user    6m 15.04s
Jun 30 08:34:12 <[g2]> sys     1m 52.94s
Jun 30 08:34:20 <Tiersten> I'd better reset the clock anyway
Jun 30 08:35:37 <jacques> [g2], that's fatslug?
Jun 30 08:35:41 <[g2]> nod
Jun 30 08:35:42 <ByronT> Frank is booting
Jun 30 08:35:56 <Tiersten> Cool
Jun 30 08:36:22 <[g2]> real    14m 44.69s
Jun 30 08:36:22 <[g2]> user    11m 35.12s
Jun 30 08:36:38 <ByronT> got the strangled beep
Jun 30 08:36:50 <[g2]> old 14m 44, new 8m 13
Jun 30 08:37:44 <Tiersten> Not 1/2 but then again we've not increased the bus speed
Jun 30 08:37:52 <Tiersten> Still a nice improvement though :)
Jun 30 08:38:15 <[g2]> oh yeah...
Jun 30 08:38:24 <jacques> nobody should expect 2x increase for anything other than dhrystone type thing
Jun 30 08:38:31 <Tiersten> yeah
Jun 30 08:38:50 <Tiersten> Unless you can wedge your entire prog into the cache and not have anything interrupt it, you won't get the full speed up
Jun 30 08:38:59 <ByronT> woot!
Jun 30 08:39:16 <ByronT> # cat /proc/cpuinfo
Jun 30 08:39:16 <ByronT> Processor : XScale-IXP425/IXC1100 rev 1 (v5b)
Jun 30 08:39:16 <ByronT> BogoMIPS : 263.78
Jun 30 08:39:16 <ByronT> Features : swp half thumb fastmult edsp
Jun 30 08:39:29 <jacques> ByronT, you must be running a 2.4 kernel
Jun 30 08:39:39 <NAiL> How many speedslugs do we have now?
Jun 30 08:39:43 <ByronT> FrankenSlug is unslung 5.5
Jun 30 08:39:49 <Tiersten> NAiL: 11
Jun 30 08:39:54 <ByronT> (for the time being)
Jun 30 08:39:57 <NAiL> It's growing fast I see ;)
Jun 30 08:39:57 <jacques> ByronT, yeah, 2.4 loses about 3 bogomips
Jun 30 08:40:05 <NAiL> I'll add another one tomorrow
Jun 30 08:41:14 <ByronT> FrankenSlug is supposed to be OpenSlug but I had to reflash to test flashing with OS X
Jun 30 08:41:30 <ByronT> popping open ProdSlug now...
Jun 30 08:41:45 <Tiersten> We've got a prog for OSX or you talking about RedBoot?
Jun 30 08:41:52 <VoodooZ_Work> does that mean I'll be #13!
Jun 30 08:42:42 <ByronT> Tiersten, I was really trying to guage how hard it was to catch the redboot telnet window with OS X
Jun 30 08:42:50 <ByronT> (a miserable experience)
Jun 30 08:43:13 <Tiersten> ah k
Jun 30 08:43:37 <ByronT> but.. we have a prog for OS X now?!?!?!?!
Jun 30 08:44:31 <Tiersten> No.  I'm asking whether you had one
Jun 30 08:45:09 <ByronT> oh, no I was doing it the "near impossible" method
Jun 30 08:46:01 <Tiersten> ah
Jun 30 08:51:38 <ByronT> oh darn it, I must have unplugged my soldering iron.. now I have to wait for it to heat back up
Jun 30 08:52:27 <Tiersten> VoodooZ_Work: Quick!  You've got a chance! :)
Jun 30 08:54:14 <ByronT> beewoolie!!!!!!!!!!!!!!!!!!!!!1
Jun 30 08:56:28 <VoodooZ_Work> hehehe. I'm rollerblading to home right now. my SlugBot will be o/c real soon!
Jun 30 08:56:30 <VoodooZ_Work> later
Jun 30 08:56:42 <Tiersten> right now?  and using IRC at the same time?  cool ;)
Jun 30 08:57:53 <beewoolie-afk> Hey folks
Jun 30 08:58:13 <beewoolie-afk> I think we've made a significant step forward.
Jun 30 08:58:26 <NAiL> Oh yes :)
Jun 30 08:58:33 <ByronT> you need to fix your entry in the yahoo db
Jun 30 08:58:36 <NAiL> beewoolie-afk: We're up to 11 slugs modified, for one ;)
Jun 30 08:58:37 <beewoolie-afk> And I don't consider this overclocking.
Jun 30 08:58:51 <beewoolie-afk> It's un-underclocking.  de-underclocking?
Jun 30 08:59:03 <ByronT> to work on my ProdSlug now
Jun 30 08:59:26 <NAiL> man.. there's heaps of backlog to read again.I haven't been gone that long
Jun 30 09:01:57 <ByronT> ProdSlug booting
Jun 30 09:02:13 <beewoolie-afk> go go go go go ...
Jun 30 09:02:26 <Tiersten> beewoolie-afk: We're up to 12 now
Jun 30 09:02:40 * beewoolie-afk rings the bell
Jun 30 09:02:50 <NAiL> This is *great*
Jun 30 09:02:57 <NAiL> ~hail beewoolie-afk
Jun 30 09:02:57 * purl bows down to beewoolie-afk and chants, "I'M NOT WORTHY!!"
Jun 30 09:03:06 <NAiL> And all you other guys ;-)
Jun 30 09:03:12 <beewoolie-afk> Had anyone determine whether or not we're going to have heat problems?
Jun 30 09:03:24 <NAiL> beewoolie-afk: Top temp so far has been 47C, afaik
Jun 30 09:03:34 <beewoolie-afk> Don't praise me, it's dyoung who did the hard part.
Jun 30 09:03:36 * Tiersten awaits mass global warming due to the increased heat from the all the NSLU2s :)
Jun 30 09:03:43 <NAiL> Some spec sheet says it can operate up to 113C within the case.
Jun 30 09:03:44 <ByronT> ProdSlug
Jun 30 09:03:45 <ByronT> # cat /proc/cpuinfo
Jun 30 09:03:45 <ByronT> Processor : XScale-IXP425/IXC1100 rev 1 (v5b)
Jun 30 09:03:45 <ByronT> BogoMIPS : 263.78
Jun 30 09:03:45 <ByronT> Features : swp half thumb fastmult edsp
Jun 30 09:03:50 <ByronT> woot
Jun 30 09:03:50 <Tiersten> beewoolie-afk: I don't believe we have any problems
Jun 30 09:03:56 <NAiL> ByronT: Unslung.
Jun 30 09:04:08 <ByronT> yeah, both of my slugs are unslung ATM
Jun 30 09:04:09 <NAiL> I.e, you're running a 2.4 kernel.
Jun 30 09:04:15 <beewoolie-afk> I ask because I'm still wondering why they would have underclocked the device.
Jun 30 09:04:20 <NAiL> 2.4 kernels leak 3Bogomips ;)
Jun 30 09:04:45 <jacques> beewoolie-afk, we can only speculate on that
Jun 30 09:04:47 <beewoolie-afk> Is this simply some sort of marketing decision?
Jun 30 09:04:55 <NAiL> beewoolie-afk: Yeah, no problems so far though. At 100% cpu, the highest measured temp was 47C. Ambient temp 29C.
Jun 30 09:05:03 <beewoolie-afk> ATM, I don't even have anything plausible.
Jun 30 09:05:10 <ByronT> you're too slow VoodooZ!!!
Jun 30 09:05:14 <beewoolie-afk> Hmm. 47C should be fine.
Jun 30 09:05:15 <ByronT> or 13
Jun 30 09:05:21 <VoodooZ> I'm home. soldering gun warming up.
Jun 30 09:05:25 <VoodooZ> damn!
Jun 30 09:05:44 <NAiL> beewoolie-afk: My guess is that they figured they'de be on the safe side. The performance difference is unlikely to be noticeable in the original firmware :(
Jun 30 09:06:04 <NAiL> VoodooZ: You're getting #13
Jun 30 09:06:14 <NAiL> It's a lucky number :D
Jun 30 09:06:25 * NAiL orders a AMD64 Dev box
Jun 30 09:06:52 <beewoolie-afk> Someone's going CRAZY from the heat.
Jun 30 09:07:08 <Tiersten> soldering gun?
Jun 30 09:07:18 <Tiersten> uh.  you mean soldering iron yes?
Jun 30 09:07:22 <ByronT> well, ProdSlug's resistor is gone for good
Jun 30 09:07:44 <jacques> ByronT, they are tiny aren't they
Jun 30 09:07:45 <ByronT> when I flicked it off, it hit the grayish carpet floor...
Jun 30 09:07:53 <Tiersten> heh
Jun 30 09:08:17 <ByronT> I found more small particles of trash than that resistor
Jun 30 09:08:41 <Tiersten> The resistor isn't that small
Jun 30 09:08:44 <beewoolie-afk> NAiL: which model?  A 939 at least?
Jun 30 09:08:50 <NAiL> beewoolie-afk: yup
Jun 30 09:08:52 <beewoolie-afk> I call 2mm small.
Jun 30 09:08:54 <VoodooZ> done.
Jun 30 09:09:00 <ByronT> it's smaller than the pieces of trash on the floor
Jun 30 09:09:01 <Tiersten> You should try overclocking a mac mini
Jun 30 09:09:03 <NAiL> 3700+, 1MB cache.
Jun 30 09:09:19 <Tiersten> The resistors for that are tiny
Jun 30 09:09:23 <beewoolie-afk> Does the mini have smaller resistors?
Jun 30 09:09:32 <Tiersten> http://www.lbodnar.dsl.pipex.com/macmini/macmini_removed_jumpers.jpg
Jun 30 09:09:33 <NAiL> of course. It's a mini :P
Jun 30 09:09:36 <beewoolie-afk> How is that possible? The electrons go single file?
Jun 30 09:09:53 <VoodooZ> booting fine...
Jun 30 09:10:11 <Tiersten> The ones on the NSLU2 aren't the smallest size.  I can't remember what the size number was for the mac mini ones
Jun 30 09:10:13 <VoodooZ> 266!!!
Jun 30 09:10:15 <VoodooZ> yipee!
Jun 30 09:10:17 <beewoolie-afk> Tiersten: holy scrit!
Jun 30 09:10:23 * beewoolie-afk rings the bell
Jun 30 09:10:26 <NAiL> VoodooZ: Welcome to the club :)
Jun 30 09:11:08 <VoodooZ> :)
Jun 30 09:11:33 <jacques> this reminds me of the tuxscreen
Jun 30 09:11:46 <jacques> stock they were running at 103MHz
Jun 30 09:12:00 <jacques> we made them run at 206MHz (stronarm)
Jun 30 09:12:06 <jacques> strongarm
Jun 30 09:12:11 <jacques> all the same things were said
Jun 30 09:12:21 <jacques> why did they come that way, they'll blow up
Jun 30 09:12:32 <jacques> we ran them for years with no ill effects at 206MHz
Jun 30 09:12:39 <jacques> two of mine are running right now
Jun 30 09:12:51 <jacques> we still dunno why they were running half-speed from the factory
Jun 30 09:12:55 <beewoolie-afk> Power consumption?
Jun 30 09:13:14 <jacques> compared to the color lcd screen?
Jun 30 09:13:17 <beewoolie-afk> AFAICT, the power argument is bogus.
Jun 30 09:13:22 <VoodooZ> number #13 better be lucky!
Jun 30 09:13:25 <beewoolie-afk> But, I still wonder.
Jun 30 09:13:31 <NAiL> beewoolie-afk: Afaik, the slug goes from ~35mA, to less than 70mA
Jun 30 09:13:42 <beewoolie-afk> NAiL: when?
Jun 30 09:13:44 <NAiL> Probably a good deal less than 70
Jun 30 09:13:54 <NAiL> when de-underclocking
Jun 30 09:14:08 <beewoolie-afk> Did you measure it?
Jun 30 09:14:20 <NAiL> ep1220 did, IIRC
Jun 30 09:14:31 <beewoolie-afk> Ah, so he's been on.  Nice.
Jun 30 09:14:34 <NAiL> Not an exact measurement
Jun 30 09:14:45 <VoodooZ> not bad of an increase: from 6 to 11 frames/sec!!
Jun 30 09:14:48 <ep1220> I did not measure at 266, 133 only
Jun 30 09:15:01 <jacques> voodooZ not bad at all!!
Jun 30 09:15:07 <NAiL> The entire slug used 35mA. But only the core doubles power usage. So it's guaranteed to be less than 70mA
Jun 30 09:15:07 <jacques> couldn't expect better IMHO
Jun 30 09:15:08 <Tiersten> Yeah.  It won't be a doubling in power consumption because the external interfaces are going to suck a fair bit of juice and they're running at exactly the same speed.
Jun 30 09:15:09 <beewoolie-afk> It should be measurable at the input, no?
Jun 30 09:15:25 <beewoolie-afk> Ethernet is always a hog.
Jun 30 09:15:37 <Tiersten> Yeah.  The PHY on mine is always really hot
Jun 30 09:15:41 <VoodooZ> yeah, I'll have to measure the current demand as I run my bot on batteries so It's important to know.
Jun 30 09:16:01 <beewoolie-afk> Oh, I forgot about the battery backup.  :-)
Jun 30 09:16:02 <VoodooZ> my heatsinks are not too hot yet and it's really hot outside.
Jun 30 09:16:02 <ep1220> BTW: It can not double, the core only consumes 50% of what the XScale needs, the other 50% go to IO
Jun 30 09:16:07 <Tiersten> you could always install a turbo switch ala PCs from about 10 years go :)
Jun 30 09:16:13 <VoodooZ> hehe
Jun 30 09:16:35 <jacques> BTW it should be possible to speed scale in software
Jun 30 09:16:43 <VoodooZ> that would be cool.
Jun 30 09:16:55 <Tiersten> Is that supported in the IXP420?
Jun 30 09:16:57 <jacques> back in the day, there was code in the arm kernel
Jun 30 09:16:59 <beewoolie-afk> jaques: How?
Jun 30 09:17:03 <Tiersten> cpufreq
Jun 30 09:17:06 <jacques> Tiersten, I dunno - it is by the pxa
Jun 30 09:17:25 <jacques> but that code was removed a while back
Jun 30 09:17:25 <beewoolie-afk> But the code doesn't allow for changing the core freq while running.
Jun 30 09:17:36 <beewoolie-afk> s/code/core/
Jun 30 09:17:45 <jacques> beewoolie-afk, on my ipaq I used to be able to echo 11 > /proc/cpu/scale
Jun 30 09:18:14 <jacques> actualyl 12 would go to 236MHz
Jun 30 09:18:29 <jacques> 10 was 206, lower values would underclock it
Jun 30 09:18:35 <beewoolie-afk> I believe that.  When we researched this de-underclocking solution, we found that the register that controls clock speed is RO once the machine is running.
Jun 30 09:18:59 <jacques> right now, I can use an app on my pxa ipaqs, (under wince and closed source)  I can change the speeds of core, proc bus, and ram bus on the fly
Jun 30 09:19:50 <beewoolie-afk> pxa != ixp
Jun 30 09:19:53 <jacques> I was assuming it was an xscale core feature, but it may be a pxa feature in which case it doesn't apply to us
Jun 30 09:20:02 <jacques> beewoolie-afk, yes I know that
Jun 30 09:21:25 <beewoolie-afk> :-)
Jun 30 09:22:18 <jacques> I guess there's no business case for dynamic clock scaling in the ixp line
Jun 30 09:22:59 <VoodooZ> i find the NEC USB chip gets hotter than the core.
Jun 30 09:23:13 <VoodooZ> And I've got big heat-sinks on both.
Jun 30 09:23:26 <jacques> does the usb chip handle the usb power ?
Jun 30 09:23:36 <jacques> if so that might explain it
Jun 30 09:23:40 <Tiersten> No
Jun 30 09:24:10 <VoodooZ> in my case yes, I have a USB2.0 camera running full blast and an unused USB2.0 flash disk
Jun 30 09:25:19 <[g2]> hey beewoolie-afk !
Jun 30 09:25:19 <VoodooZ> seems stable. load is up to 2.0 and it's not that hot.
Jun 30 09:25:39 <beewoolie-afk> [g2]: word up
Jun 30 09:25:51 <beewoolie-afk> I think we're at 14 de-underclockers
Jun 30 09:26:17 <VoodooZ> 13 to be exact
Jun 30 09:26:21 <NAiL> VoodooZ: I ran "cat /dev/zero | gzip - | gunzip - > /dev/null" for about an hour. It was still comfortable to touch.
Jun 30 09:26:34 <VoodooZ> cool. (pun intended)
Jun 30 09:26:36 <NAiL> Aren't like.. three of 'em rwhitby?
Jun 30 09:26:50 <[g2]> yeah, and I've got 2
Jun 30 09:26:53 <VoodooZ> yeah, and [g2]s
Jun 30 09:26:59 <beewoolie-afk> Did some of you folks put heat sinks on the chips in the slug?
Jun 30 09:27:00 <[g2]> what he said
Jun 30 09:27:02 <VoodooZ> Byron has 2
Jun 30 09:27:07 <NAiL> I'm getting another tomorrow :D
Jun 30 09:27:07 <[g2]> nope
Jun 30 09:27:11 <VoodooZ> I did way back last year. 
Jun 30 09:27:21 <VoodooZ> I have no idea why.
Jun 30 09:27:24 <jacques> heh
Jun 30 09:27:26 <VoodooZ> but now I do
Jun 30 09:27:35 <[g2]> my avila didn't have a heat sync and the runs 266
Jun 30 09:27:44 <beewoolie-afk> So, no one would be saying thst heatsinks are necessary.
Jun 30 09:28:03 <[g2]> depends on the app and environs
Jun 30 09:28:34 <beewoolie-afk> ...on the slug....
Jun 30 09:28:35 <VoodooZ> I have lots of them so didn't cost me much.
Jun 30 09:28:55 <VoodooZ> makes the SlugBot look meaner
Jun 30 09:28:58 <[g2]> beewoolie-afk, I meant is you slug in you AC'd house or in your car ?
Jun 30 09:29:49 <[g2]> it's it packed in a closet or in a wide open space
Jun 30 09:30:01 <beewoolie-afk> [g2]: My slug is in a un-airconditioned closet with other gear.  It gets hot in there when the ambient temp rises to the 20's.
Jun 30 09:30:36 <[g2]> dyoung-zzzz, was testing with a cd-case over top an it only went to 48C
Jun 30 09:30:50 <ByronT> wow, I found the flicked away resistor
Jun 30 09:32:08 <[g2]> ahh... jacques I wonder if the alignment handling is any different between 11.2 and 12
Jun 30 09:33:29 <VoodooZ> I soldered one side of the resistor back on the pad so I wouldn't lose it.
Jun 30 09:34:08 <ByronT> though I really don't know why I looked and/or care....
Jun 30 09:35:42 <Tiersten> One day you might be thankful that you've got a SMD resistor handy.  Unlikely however :)
Jun 30 09:36:06 <VoodooZ> unless you don't remember the value that was there before. Again, unlikely.
Jun 30 09:36:14 <ByronT> if I would be able to re-track down where I placed that resistor
Jun 30 09:36:36 <jacques> [g2], hmmmmmm
Jun 30 09:36:37 <VoodooZ> Thus why i soldered mine on one of the pads.
Jun 30 09:37:03 <Tiersten> ByronT: You found it, put it in a safe place and have now lost it??
Jun 30 09:37:21 <ByronT> no, I'm just anticipating the next time I lose it
Jun 30 09:37:24 <Tiersten> ah
Jun 30 09:37:31 <Tiersten> I just taped mine to the inside of the case
Jun 30 09:37:53 <ByronT> and a "safe place" is an undefined variable at my office
Jun 30 09:38:05 <ByronT> (the 2 year old comes and visits from time to time)
Jun 30 09:38:18 <Tiersten> Tape it to the NSLU2 or tombstone it on one of the pads
Jun 30 09:38:52 <[g2]> heh... tombstone  ... haven't heard that phrase for a while
Jun 30 09:39:10 <[g2]> where's the wave-flow solding station when you need it
Jun 30 09:39:17 <Tiersten> :)
Jun 30 09:39:28 <[g2]> LoL
Jun 30 09:49:48 <[g2]> Jacmet, 
Jun 30 09:50:01 <[g2]> doh!... damn autocomplete
Jun 30 09:50:04 <[g2]> jacques, 
Jun 30 09:50:15 <[g2]>         Everything is up to date. Type 'make test' to run test suite.
Jun 30 09:50:16 <[g2]> real    1h 1m 45s
Jun 30 09:50:16 <[g2]> user    58m 51.91s
Jun 30 09:50:24 <[g2]> perl build on TurboFatty
Jun 30 09:51:05 <jacques> [g2], interesting
Jun 30 09:51:12 <beewoolie-afk> Do we have confirmation on the speed increase?
Jun 30 09:51:43 <[g2]> oh we've had lots of confirmations
Jun 30 09:51:50 <jacques> <jacques> my perl 'make' time went from real    89m2.520s to real    61m26.581s
Jun 30 09:52:14 <VoodooZ> My CV test is far from synthetic so I think it's pretty f'n real
Jun 30 09:52:28 <beewoolie-afk> 50% improvement.  Not bad.  Anyone been able to check the network throughput?
Jun 30 09:52:57 <VoodooZ> USB2.0 in high speed mode didn't change as expected.
Jun 30 09:52:58 <[g2]> It'd imagine with the interrupt driven driver it'd be a bunch faster
Jun 30 09:53:00 <jacques> beewoolie-afk, AFAIK not yet
Jun 30 09:55:00 <beewoolie-afk> VoodooZ: I wouldn't expect USB to change much.  IIRC, the bottleneck on USB is the controller itself which is half-duplex.
Jun 30 09:55:03 <jacques> [g2], I bet your python test is tons faster than mine - so much thrashing
Jun 30 09:55:18 <jacques> I think he meant he didn't expect usb to change
Jun 30 09:55:22 <VoodooZ> nod
Jun 30 09:55:25 <jacques> and since pci bus hasn't changed...
Jun 30 09:55:51 <VoodooZ> now, if I could overclock the memory! ;)
Jun 30 09:56:24 <jacques> I don't see that in the cards :-\
Jun 30 09:56:28 <beewoolie-afk> Ah, a puntuation mishap.
Jun 30 09:56:30 <VoodooZ> hehe
Jun 30 09:57:03 <VoodooZ> We've pretty much upgraded/overclocked everything else! (i.e. RAM, FLASH, Core...)
Jun 30 09:57:09 <jacques> even if you could get the proc to comply, the ram or traces might not
Jun 30 09:57:48 <VoodooZ> I would try to overclock the core before that anyways. 333Mhz anyone?
Jun 30 09:57:50 * [g2] kicks off python tests
Jun 30 09:57:54 <VoodooZ> :)
Jun 30 09:58:05 <VoodooZ> anyways, I'm heading back to work now...
Jun 30 09:58:23 <beewoolie-afk> Thanks for stopping by.
Jun 30 09:58:23 <jacques> [g2], my python tests thrashed so much, they were on;y three minutes faster (80 vs 83) on turbo
Jun 30 09:58:29 <VoodooZ> see you in few minutes. (skating uphill :( )
Jun 30 09:58:31 <[g2]> jacques, the old python test took 35m 50secs
Jun 30 09:58:50 <jacques> [g2], damn, that extra ram does help in that case don't it
Jun 30 09:58:57 <[g2]> nod
Jun 30 09:59:06 <[g2]> that's a swap thing
Jun 30 09:59:24 <jacques> once you start thrashing, all bets are off
Jun 30 09:59:40 <[g2]> you just thrash faster :)
Jun 30 10:01:46 <[g2]> jacques, I think it's actually less than the 35 min
Jun 30 10:02:01 <[g2]> cause some tests started to be run for the second time
Jun 30 10:02:25 <jacques> yeah it runs twice, second time compiled
Jun 30 10:03:33 <[g2]> ah.. then that's a shortened run
Jun 30 10:03:52 <jacques> one would hope
Jun 30 10:06:57 <[g2]> jacques, I meant that my 35m time was short because it was 1 full pass + 50ish tests
Jun 30 10:07:34 <jacques> ah
Jun 30 10:08:13 <jacques> at the end of each phase the python process gets to 53MB
Jun 30 10:08:24 <jacques> which is too much for a stock slug
Jun 30 10:11:34 <beewoolie-afk> Python is notoriously bad with memory.  It can be optimized by those willing to delve.
Jun 30 10:11:50 <beewoolie-afk> Infamous, in fact.
Jun 30 10:13:38 <jacques> I think they just never imagined someone would run the testsuite on a < 64MB machine
Jun 30 10:13:59 <jacques> if they just started a new python process for each test it would be fine
Jun 30 10:14:40 <jacques> beewoolie-afk, it sounds like they just don't think about memory usage much :-)
Jun 30 10:15:06 <Tiersten> Did anybody try > 266MHz?
Jun 30 10:15:14 <Tiersten> Yes.  I know the part isn't rated for that
Jun 30 10:15:19 <beewoolie-afk> I don't know much about it in detail.  I do know that bitbake was so memory hungry that many people couldn't run it.
Jun 30 10:15:29 <[g2]> can't do, as rwhitby-asleep said the speed grade if fused
Jun 30 10:15:33 <[g2]> can't do, as rwhitby-asleep said the speed grade is fused
Jun 30 10:15:35 <beewoolie-afk> Tiersten: How would we do it?  Change the crystal?
Jun 30 10:15:53 <[g2]> as [g2] said we could OC the FSB :)
Jun 30 10:16:11 <Tiersten> ah k
Jun 30 10:16:24 <beewoolie-afk> jacques: I also know that someone made changes to bbake and reduced the memory usage significantly.
Jun 30 10:16:49 <VoodooZ_Work> o/c the fsb is one way yes. That's what I do with PCs.
Jun 30 10:17:05 <VoodooZ_Work> I'm sure this part does 300+ easily. But probably not the rest.
Jun 30 10:17:15 <[g2]> but that would ripple through
Jun 30 10:17:25 <jacques> beewoolie-afk, yeah it went from 450MB to 15MB
Jun 30 10:17:29 <VoodooZ_Work> I don't think the clock divider are very flexible though. /2 ? /3?
Jun 30 10:17:57 <jacques> I think the pci would fail first, followed closedly by the ram
Jun 30 10:18:02 <beewoolie-afk> jacques: I think it started at more than that.
Jun 30 10:18:09 <jacques> maybe even the flash
Jun 30 10:18:14 <beewoolie-afk> I pretty satisfied with 266.
Jun 30 10:18:36 <Tiersten> So long as we don't go wildly out of spec, the NEC chip should be able to handle it
Jun 30 10:18:44 <VoodooZ_Work> i doubt they are individually clock locked like PCs too
Jun 30 10:18:52 <jacques> from my reading, the NEC usb chips (especially the older models like this one) don't overclock well at all
Jun 30 10:19:00 <Tiersten> Okay.  I guess not :)
Jun 30 10:19:17 <VoodooZ_Work> satisfied? Nobody is satisfied in here. That's why we hacked that slug in the first place! :)
Jun 30 10:19:21 <jacques> it's weird - some PCI devices can do 66MHz no problem, and some can't do 37MHz
Jun 30 10:19:24 <Tiersten> I wouldn't touch NEC USB chips with a barge pole now.  The one in the NSLU2 has put me off
Jun 30 10:19:37 <Tiersten> Buggy slow POS
Jun 30 10:19:42 <beewoolie-afk> VoodooZ_Work: speak for yourself, hombre.
Jun 30 10:19:45 <jacques> Tiersten, that agrees with my reading :-)
Jun 30 10:19:47 <VoodooZ_Work> hehehe
Jun 30 10:19:52 <beewoolie-afk> Tiersten: hear hear.
Jun 30 10:20:22 <beewoolie-afk> I notice that most ther NAS devices, like the slug sort, don't have USB.
Jun 30 10:20:29 <beewoolie-afk> Or, don't have use 2.0
Jun 30 10:20:59 <jacques> there is a later version of the nec chip that's supposed to be better - can do full duplex
Jun 30 10:21:10 <beewoolie-afk> Pin compatible...?
Jun 30 10:21:21 <jacques> yah
Jun 30 10:21:27 <beewoolie-afk> Hmm.
Jun 30 10:21:37 * beewoolie-afk is not longer satisfied.
Jun 30 10:21:41 <Tiersten> The Terastation/Linkstation has the NEC USB chip as well.  It's a newer revision however
Jun 30 10:21:42 <jacques> I read all this years ago when I needed a usb2.0 card supported by linux
Jun 30 10:21:54 <VoodooZ_Work> hehe
Jun 30 10:22:19 <jacques> the one I have in this machine is the later rev
Jun 30 10:25:43 <jacques> where are the big board photos? I've lost them :-(
Jun 30 10:26:07 <VoodooZ_Work> under info i think
Jun 30 10:26:45 <Tiersten> The freq locking, this for certain?
Jun 30 10:27:52 <beewoolie-afk> Tiersten: You could try to put the resistor on R81 (IIRC) and seeif you get a boost.
Jun 30 10:28:02 <Tiersten> I've removed R83 though
Jun 30 10:28:12 <beewoolie-afk> Hmm.  Why?
Jun 30 10:28:39 <Tiersten> To get 266MHz
Jun 30 10:29:18 <beewoolie-afk> All you should need to do is remove R64.
Jun 30 10:29:36 <Tiersten> The "official" method was to remove R83
Jun 30 10:29:44 <jacques> beewoolie-afk, the official method is now remove R83
Jun 30 10:29:45 <beewoolie-afk> Really?  that isn't what I did.
Jun 30 10:29:50 <beewoolie-afk> Why?
Jun 30 10:30:12 <jacques> one reason because it's easier/less dangerous
Jun 30 10:30:26 <beewoolie-afk> Huh?  Dangerouse how?  Terrorist attack?
Jun 30 10:30:52 <jacques> someone else explain it
Jun 30 10:31:10 <beewoolie-afk> Is it on the wiki?
Jun 30 10:31:21 <Tiersten> http://www.nslu2-linux.org/wiki/HowTo/OverClockTheSlug
Jun 30 10:31:27 <Tiersten> That just has the steps
Jun 30 10:31:32 <Tiersten> The reason why is in the logs
Jun 30 10:31:43 <jacques> less dangerous in that there's less that can be damaged close to R83
Jun 30 10:32:17 <beewoolie-afk> whatever.  I don't see a difference.
Jun 30 10:32:31 <beewoolie-afk> At least r64 is further from other solder pads.
Jun 30 10:33:25 <beewoolie-afk> Removing R83 makes the clock divider value 0b011?
Jun 30 10:34:15 <jacques> 011 is what I've seen mentioned yeah
Jun 30 10:34:26 <beewoolie-afk> Hmm.  that doesn't make sense to me.  which bit is R83?
Jun 30 10:34:40 <beewoolie-afk> I thought that removing the resistors would make the bits 0, not 1.
Jun 30 10:35:00 <VoodooZ_Work> I wonder what would happen if I swapped the 33Mhz crystal for a 40Mhz one. <Evil Laugh>
Jun 30 10:35:56 <jacques> beewoolie-afk, yeah that agrees with what I though the value should end up being - 100  but I'm just repeating what I've seen others say about that
Jun 30 10:37:42 <jacques> I think [g2] said he had dumped the values so he has first-hand knowledge
Jun 30 10:38:03 <jacques> VoodooZ_Work, that could possibly work, I'd like to hear all about it if you try :-)
Jun 30 10:38:05 <[g2]> yes the old value was 41
Jun 30 10:38:25 <VoodooZ_Work> jacques: hehehe I was hoping someone would try for me :)
Jun 30 10:38:46 <VoodooZ_Work> I wonder if linux would need to be tweaked though?
Jun 30 10:40:32 <VoodooZ_Work> Isn't there software/hardware expecting a specific clock? or is it all synchronous?
Jun 30 10:41:03 <[g2]> Jun 29 17:07:18 <[g2]>  dump 0xc4000020
Jun 30 10:41:03 <[g2]> Jun 29 17:07:18 <[g2]>  c4000020: 00 81 ff ee
Jun 30 10:41:08 <[g2]> that's the avila
Jun 30 10:41:15 <[g2]> running @ 266
Jun 30 10:41:22 <beewoolie-afk> jacques: Well, I suppose that R64 may not be A23.
Jun 30 10:41:32 <[g2]> Jun 29 16:27:25 <[g2]>  c4000020: 00 41 ff ee
Jun 30 10:41:45 <[g2]> that's slug
Jun 30 10:41:51 <[g2]> @ 133
Jun 30 10:41:58 <beewoolie-afk> [g2]: IIRC, dyoung said that R64 was A23, R83 was A22 and R81 was A21.  does this still stand?
Jun 30 10:42:02 <[g2]> beewoolie-afk, you can just dump the value with APEX
Jun 30 10:42:17 <beewoolie-afk> I don't have apex on that machine and I don't have time to look at it today.
Jun 30 10:43:48 <[g2]> jacques, 2 passes of python on TurboFatty 
Jun 30 10:43:51 <[g2]> real    41m 38.74s
Jun 30 10:45:38 <[g2]>  dump 0xc4000020
Jun 30 10:45:38 <[g2]> c4000020: 00 61 ff ee
Jun 30 10:45:39 <VoodooZ_Work> There's 3 crystals on the board. One small 32Khz one for the RTC, and two big cans.
Jun 30 10:45:41 <beewoolie-afk> I tried to dump the register value using Redboot, but it reset.
Jun 30 10:46:03 <VoodooZ_Work> I can't read the value off the big one near the Ethernet jack from the pixs.
Jun 30 10:46:05 <[g2]> yeah Redboot resets for me
Jun 30 10:46:08 <beewoolie-afk> [g2]: Whoa!  so, R83 is the A23 line?
Jun 30 10:46:27 <[g2]> what's the wiki say ?
Jun 30 10:46:31 <beewoolie-afk> No, I got that wrong.
Jun 30 10:46:49 <Tiersten> R83 is whatever pin 3 on the flash is
Jun 30 10:46:54 <beewoolie-afk> R83 is A21 and removing the resistor lets the bit float high.
Jun 30 10:46:59 <jacques> [g2], esxcellent - that's 1/2 my time
Jun 30 10:47:32 <[g2]> I've been wanting to run some Apache/php5 tests
Jun 30 10:47:33 <beewoolie-afk> Seems like we should have some explanation of this on the wiki.  I'd do it, but I'm not confident I have the data right.
Jun 30 10:56:14 <[g2]> what's that intel pub number with the boot strap options ?
Jun 30 10:56:32 <beewoolie-afk> [g2]: Do you mean the clock controlbits?
Jun 30 10:56:35 * [g2] has way too many dirs and Intel pdfs.... 
Jun 30 10:56:38 <[g2]> yeah
Jun 30 10:56:40 <beewoolie-afk> It's in the 2005 doc.
Jun 30 10:56:44 <ep1220> g2: the newset version is in the core-dev Manual 25248005
Jun 30 10:56:45 <[g2]> the app note
Jun 30 10:57:00 <ep1220> THe AppNote is 6 months older, and slightly different
Jun 30 10:57:04 <beewoolie-afk> ixp42x-guide-25248005.pdf
Jun 30 10:58:07 <[g2]> THX guys
Jun 30 11:01:05 <[g2]> Hmmm R83 goes to the flash
Jun 30 11:01:15 <[g2]> ?
Jun 30 11:01:27 <beewoolie-afk> I think it should.  All of them should.
Jun 30 11:01:31 <ep1220> THe FLASH also needs address lines
Jun 30 11:01:33 <beewoolie-afk> ...
Jun 30 11:01:53 <beewoolie-afk> Except that the flash is only 8MB.
Jun 30 11:01:57 <[g2]> I wonder how this will effect shutdown -r
Jun 30 11:02:04 <beewoolie-afk> R64 probably doesn't need to go to flash.
Jun 30 11:02:10 <beewoolie-afk> The rest do.
Jun 30 11:02:16 <[g2]> from the eternal reset
Jun 30 11:05:59 <VoodooZ_Work> Is it me or the Crystal on the board says it's 30Mhz?
Jun 30 11:06:00 <ep1220> VoodooZ_Work: the crystal close to the network chip is 25MHz
Jun 30 11:06:17 <VoodooZ_Work> yeah, that probably is the 24Mhz clock for USB 2.0 chip
Jun 30 11:06:38 <VoodooZ_Work> Unless I'm mixed up with USB1.0
Jun 30 11:06:42 <ep1220> At the bottom of the board is a 33MhZ oscilaltor which drives the CPU
Jun 30 11:06:50 <VoodooZ_Work> is it really 33?
Jun 30 11:06:59 <VoodooZ_Work> The drawing says 30.00.
Jun 30 11:07:00 <ep1220> yes
Jun 30 11:07:03 <VoodooZ_Work> ok.
Jun 30 11:07:45 <VoodooZ_Work> if that that crystal drives the PCI bus, then FSB is PCI * 4 = 133Mhz. ANd the core runs at FSB * 2 = 266Mhz.
Jun 30 11:07:50 <VoodooZ_Work> DOes that make sense?
Jun 30 11:08:25 <ep1220> Not sure if it dirves PCI directly, as PCI can be set to 66 as well
Jun 30 11:08:39 <VoodooZ_Work> So if I substitute the 33Mhz xtal for something like 40Mhz I would get: PCI = 40Mhz, FSB = 160Mhz & Core = 320MHz. ???
Jun 30 11:08:47 <ep1220> Yes
Jun 30 11:08:56 <ep1220> BUt it might well be the PLL would not lock at all
Jun 30 11:09:04 <VoodooZ_Work> there's a multiplier between the crystal and the PCI bus then?
Jun 30 11:09:12 <VoodooZ_Work> yeah, that's true.
Jun 30 11:09:28 <VoodooZ_Work> Unless the PLL can be "tweaked"?
Jun 30 11:09:40 <VoodooZ_Work> What's the PLL chip like? and where is it?
Jun 30 11:10:14 <ep1220> It is inside the XScale.
Jun 30 11:10:31 <VoodooZ_Work> oh i see.
Jun 30 11:10:35 <ep1220> 40Mz is nearly 25% off
Jun 30 11:10:43 <VoodooZ_Work> That means it can be programmed then right?
Jun 30 11:10:54 <VoodooZ_Work> how about 37Mhz then?
Jun 30 11:11:18 <ep1220> It is programmed by the 3 bits we are fiddling around.
Jun 30 11:11:35 <ep1220> Plus some info hardcoded into the Xscale 
Jun 30 11:11:42 <VoodooZ_Work> ah, I thought those were simply dividers/mutipliers.
Jun 30 11:11:53 <VoodooZ_Work> what's our best bet then?
Jun 30 11:13:16 <VoodooZ_Work> Digikey has oscillators ranging from 33, 35, 35, 36 and 38Mhz
Jun 30 11:13:20 <beewoolie-afk> ep1220: the 25MHz crystal is for the Ethernet.
Jun 30 11:13:32 <VoodooZ_Work> ah ok
Jun 30 11:13:37 <ep1220> Must check datasheet, maybe there is some external circuit for the PLL which can be tweaked
Jun 30 11:13:53 <VoodooZ_Work> 300Mhz is surely possible.
Jun 30 11:14:33 <VoodooZ_Work> It would be nicer if we could change just the core and not mess with the PCI bus.
Jun 30 11:14:55 <VoodooZ_Work> Is the FSB for the 533Mhz IXP still 133Mhz?
Jun 30 11:16:10 <ep1220> lets do PCI first:
Jun 30 11:16:14 <VoodooZ_Work> that means the FSB of 133Mhz is simply multiplied by 2 for 266, 3 for 400 or 4 for 533.
Jun 30 11:16:30 <VoodooZ_Work> If my assumptions are ok that is.
Jun 30 11:16:42 <VoodooZ_Work> I'm no Xscale expert. I just know PCs.
Jun 30 11:17:32 <ep1220> The Dev Manual says the max Expansionbus Clock is 66MHz
Jun 30 11:17:40 <VoodooZ_Work> I know that most PCI devices in PC can cope with 37Mhz no problem.
Jun 30 11:17:55 <VoodooZ_Work> which can be configured using the same reg.
Jun 30 11:18:02 <VoodooZ_Work> in s/w
Jun 30 11:19:17 <ep1220> You can try to replace the crystal.
Jun 30 11:19:53 <VoodooZ_Work> yeah, but with what value? If I had one of each I would start with 35Mhz and go up.
Jun 30 11:20:09 <ep1220> I woudl start with the closest to 33 which You can get
Jun 30 11:20:18 <VoodooZ_Work> which is 35Mhz.
Jun 30 11:20:23 <beewoolie-afk> You guyz are MAD
Jun 30 11:20:29 <VoodooZ_Work> hihihihi
Jun 30 11:20:45 <beewoolie-afk> MAD I tell you
Jun 30 11:20:48 <VoodooZ_Work> mad is not right, we are scientifically-curious!
Jun 30 11:20:51 <VoodooZ_Work> :)
Jun 30 11:21:02 <beewoolie-afk> Scientifically MAD
Jun 30 11:21:12 <VoodooZ_Work> ok, that will do :)
Jun 30 11:21:14 <beewoolie-afk> bwahahahaha
Jun 30 11:21:19 <VoodooZ_Work> lol
Jun 30 11:21:38 <VoodooZ_Work> It's simple enough. just two pins! Not even surface mount too! 
Jun 30 11:21:46 <beewoolie-afk> Seriously, tho.  Is it worth it?  The performance gain is likely to me nominal.
Jun 30 11:22:03 <VoodooZ_Work> I know. But it would fun to try.
Jun 30 11:22:15 <beewoolie-afk> Unless you can boost the clock 50%, I wonder if you'll even be able to measure the gain.
Jun 30 11:22:22 <VoodooZ_Work> 38Mhz would be sweet.
Jun 30 11:22:24 <beewoolie-afk> Indeed.  Hence, the MAD part  :-)
Jun 30 11:22:52 <VoodooZ_Work> 38Mhz would give us around 300Mhz.
Jun 30 11:23:10 <VoodooZ_Work> and as a side-effect overclock the USB/ethernet/etc...
Jun 30 11:23:40 <beewoolie-afk> I'm not sure you can overclock the PHY.  The transmit rate is governed by the spec.
Jun 30 11:23:56 <VoodooZ_Work> not that I care anyways as my robot won't use it.
Jun 30 11:24:06 <ep1220> INtel has an Appnote on clocking Doc Number 305588
Jun 30 11:24:13 <beewoolie-afk> Hmm.
Jun 30 11:24:17 <ep1220> beewoolie-afk: thy PHY has its own clock
Jun 30 11:24:33 <ep1220> as does USB
Jun 30 11:24:42 <ep1220> So You can not overclock these Bus-Speeds
Jun 30 11:24:46 <VoodooZ_Work> but the RAM would be though
Jun 30 11:25:00 <ep1220> the RAM is always at 133Mhz for all clocks
Jun 30 11:25:02 <VoodooZ_Work> to 152Mhz
Jun 30 11:25:18 <VoodooZ_Work> isnt' that 133 based off PCI clock though?
Jun 30 11:25:41 <ep1220> yes.  33Mhz in -> 133Mhz on SDRAM
Jun 30 11:26:00 <ep1220> ( meant for all 3 official core-clocks)
Jun 30 11:26:14 <VoodooZ_Work> so 38Mhz in (x4) -> 152Mhz ???
Jun 30 11:26:39 <ep1220> yes
Jun 30 11:26:51 <VoodooZ_Work> hahahaha I'm so MAD today!
Jun 30 11:26:51 <ep1220> Intel Spec: Clock tolerance: +/- 50ppm
Jun 30 11:27:20 <ep1220> Clock should be 33.33MHz
Jun 30 11:27:54 <beewoolie-afk> ep1220: yeah, PHY has separate 25MHz clock.  Upping the core speed should make the ethernet interface faster, tho, because the MAC is synthesized by the core.
Jun 30 11:28:09 <beewoolie-afk> That's why I've been curious about ethernet performance tests.
Jun 30 11:28:36 <VoodooZ_Work> I can try a file transfer for you.
Jun 30 11:29:07 <ep1220> The clock spec means: The slug is already off by 1% and thus out of spec
Jun 30 11:29:32 <ep1220> Maybe this is the reason why they have the testpoint at the PLL-lock signal
Jun 30 11:29:40 <ep1220> and donot clock at 266
Jun 30 11:29:40 <VoodooZ_Work> Unless i submerge the whole thing in liquid nitrogen?!?! (more evil laugher)
Jun 30 11:30:23 <beewoolie-afk> Out of spec meaning that it's running faster, at 266, than the core can handle?
Jun 30 11:30:35 <ep1220> No, the clock is too slow
Jun 30 11:30:44 <ep1220> thy use 33, but should be 33,33
Jun 30 11:30:46 <VoodooZ_Work> well, same applies to their PC processor but people still overclock them right?
Jun 30 11:31:19 <ep1220> never read PC CPU clock specs
Jun 30 11:31:24 <beewoolie-afk> I don
Jun 30 11:31:37 <VoodooZ_Work> me neither. I just overclock them! :)
Jun 30 11:31:39 <beewoolie-afk> I don't see how using a slower crystal could be a problem.
Jun 30 11:32:21 <ep1220> maybe the PLL can have problems to lock ?
Jun 30 11:33:14 <beewoolie-afk> That doesn't jive with my understanding of PLL circuits.
Jun 30 11:38:38 <VoodooZ_Work> I tried cat'ing a 43Megs file from an NFS drive to /dev/null and it took 4.79secs. which equals to ~8.99Megs/sec
Jun 30 11:40:55 <beewoolie-afk> VoodooZ_Work: Unfortunately, NFS isn't a very reliable method for testing.  It's affect (?) on the page cache makes it bothersome.
Jun 30 11:41:38 <beewoolie-afk> Unfortunately, I don't have anything more reliable.  The bandwidth tester that lots of people use (I forget the name) is flaky and tends to crash on these ARM targets.  I don't know why.
Jun 30 11:41:40 <[g2]> I'm missing the page cache issue there
Jun 30 11:42:11 <ByronT-Away> good gracious, I go to lunch happy with a TurboSlug, and I come back, read the logs, and you guys are already discussing a *real* overclocking
Jun 30 11:42:14 <beewoolie-afk> When NFS reads, it fills RAM with pages.  That fill will evict code pages, even if those code pages are being used.
Jun 30 11:42:22 <beewoolie-afk> :-)
Jun 30 11:42:41 <beewoolie-afk> I found a nasty TLB bug because of this page-cache effect.
Jun 30 11:42:56 <[g2]> so the pages get swapped out ?
Jun 30 11:43:20 <beewoolie-afk> Not necessarily.  The pages may simply be discarded.  Code can be reread from the source.
Jun 30 11:43:33 <[g2]> right
Jun 30 11:43:37 <beewoolie-afk> I don't recall whether or not the kernel does it this way.
Jun 30 11:43:58 <[g2]> VoodooZ_Work, it's that UDP or TCP
Jun 30 11:43:59 <beewoolie-afk> However, I believe if a code page hasn't been modified, it may choose to discard instead of swapping.
Jun 30 11:45:11 <VoodooZ_Work> [g2]: no idea. just the default.
Jun 30 11:45:22 <[g2]> beewoolie-afk, are you saying that VoodooZ_Work just needs a file bigger than 43MB like 80MB that reduces/eliminates the chance that the pages are already there
Jun 30 11:45:27 <VoodooZ_Work> How about a pure TCP speed test using C?
Jun 30 11:45:27 <[g2]> DP
Jun 30 11:45:28 <[g2]> UDP
Jun 30 11:45:33 <[g2]> ttcp
Jun 30 11:45:41 <VoodooZ_Work> ?
Jun 30 11:46:06 <[g2]> the old slug line rate ttcp
Jun 30 11:46:10 <VoodooZ_Work> ByronT: Who's going to be first to swap their crystal? :)
Jun 30 11:46:11 <[g2]> iirc
Jun 30 11:46:41 <VoodooZ_Work> we could use netcat to pipe data via TCP
Jun 30 11:46:51 <ByronT> not I... I would have to go dig out my soldering station... and no caffeine for a few days
Jun 30 11:47:05 <VoodooZ_Work> there's a netperf package in OE
Jun 30 11:47:11 <VoodooZ_Work> hehe
Jun 30 11:47:39 <beewoolie-afk> Yeah.  Netperf is the one that doesn't work for me.
Jun 30 11:47:51 <VoodooZ_Work> If I had a 35-38Mhz crystal handy I would but I only have 40Mhz. 
Jun 30 11:48:39 <[g2]> iirc I could do about 70Mbs with TCP and write to disk
Jun 30 11:48:54 <beewoolie-afk> [g2]: That is part of the issue.  The file needs to be twice as big as RAM.  Still, it isn't the best test because NFS is so messy with RAM.  Believe it or not, I think that FTP tests are better, but I don't know if you can copy with FTP to /dev/null.
Jun 30 11:49:23 <VoodooZ_Work> netcat is the perfect thing for that
Jun 30 11:49:25 <beewoolie-afk> netcat might be a good bet.
Jun 30 11:49:28 <[g2]> ttcp is just a plain C program
Jun 30 11:49:37 <[g2]> nc another
Jun 30 11:49:39 <beewoolie-afk> We like simple.
Jun 30 11:49:45 <[g2]> but ttcp was much faster than nc iirc
Jun 30 11:50:12 <[g2]> the avila board with GigE benched ttcp > 200Mbs
Jun 30 11:50:37 <ep1220> VoodooZ_Work: You need an oscillator, not just a crystal
Jun 30 11:50:52 <VoodooZ_Work> sorry, that's what I meant.
Jun 30 11:50:55 <VoodooZ_Work> oscillator cans
Jun 30 11:51:41 <beewoolie-afk> [g2]: BTW, did you ever add samba and nfs daemon to the feed?
Jun 30 11:54:41 <VoodooZ_Work> cool. netcat (nc) is part of busybox already!
Jun 30 11:59:34 <VoodooZ_Work> using nc command and transfering a 43Meg file it took 4.163secs.
Jun 30 11:59:45 <VoodooZ_Work> that's a plain TCP socket
Jun 30 11:59:46 <[g2]> NAiL, added samba
Jun 30 12:00:01 <[g2]> I think nfs is already there
Jun 30 12:00:21 <[g2]> there's just a couple commands for a server
Jun 30 12:00:56 <[g2]> VoodooZ_Work, so nc is xferring near line rate
Jun 30 12:01:03 <VoodooZ_Work> Which gives me ~10MB/s
Jun 30 12:01:28 <beewoolie-afk> So, if I pull from the default place...looking for it...
Jun 30 12:01:39 <VoodooZ_Work> Let me try a bigger file
Jun 30 12:01:44 <jacques> so your evil overclocked slugs aren't exploding under heavy network usage eh?   interesting
Jun 30 12:02:04 <ByronT> jacques, I thought you said your case melted... ;)
Jun 30 12:02:29 <jacques> yeah it melted enough that there's a hole in it, and I can see a red glow coming from inside
Jun 30 12:02:30 <beewoolie-afk> [g2]: I don't see exportfs.
Jun 30 12:02:46 <beewoolie-afk> jacques: What melted your case?
Jun 30 12:02:57 <[g2]> you probably didn't ipkg install any nfs things
Jun 30 12:02:59 <ByronT> (it's a joke)
Jun 30 12:03:00 <[g2]> like nfs-utils
Jun 30 12:03:13 <beewoolie-afk> [g2]: I'm looking to see how it's configured.
Jun 30 12:03:37 <VoodooZ_Work> ok, sending a 86Meg file to my slug using nc took 11.077secs = 7.775MB/s
Jun 30 12:03:49 <[g2]> then look in the post_install section of nfs-utils
Jun 30 12:03:55 <beewoolie-afk> That's a reasonable speed.
Jun 30 12:04:18 <VoodooZ_Work> Can someone try it with a SluggistSlug?
Jun 30 12:04:24 <jacques> beewoolie-afk, just a running overclock heat joke :-)
Jun 30 12:04:34 <VoodooZ_Work> SluggishSlug that is
Jun 30 12:04:39 <beewoolie-afk> OK, help a pathetic old coger.  What is the name of the feed.
Jun 30 12:04:46 <beewoolie-afk> jacques: OK.  I'm gullible.
Jun 30 12:05:01 <jacques> sorry I didn't think anyone would believe it :-D
Jun 30 12:06:26 <beewoolie-afk> [g2]: I see references to an openslug package feed, but no URL.
Jun 30 12:07:45 <beewoolie-afk> [g2]: Do I have to build all of the packages myself?
Jun 30 12:07:46 <VoodooZ_Work> Strange The second time around it took around 9.352secs for 10MB/s
Jun 30 12:07:54 <[g2]> the packages are build
Jun 30 12:08:14 <[g2]> are you booted on an external disk ?
Jun 30 12:08:20 <beewoolie-afk> Yeah.
Jun 30 12:08:27 <[g2]> are you I-net connected ?
Jun 30 12:08:41 <[g2]> DNS etc..
Jun 30 12:08:42 <beewoolie-afk> Oh wait, it looks like it isn't booting to the HD.
Jun 30 12:08:47 <beewoolie-afk> yeah.  I have net connectivity.
Jun 30 12:09:10 <beewoolie-afk> I'm SSHd right now.
Jun 30 12:09:12 <[g2]> *if* you're on the hd then just "ipkg update"
Jun 30 12:09:46 <[g2]> if not, you'll error or fill up the jffs2 depending on how old your sw is
Jun 30 12:09:53 <beewoolie-afk> Don't I need to set the IPKG source?
Jun 30 12:10:02 <[g2]> it's already set 
Jun 30 12:10:20 <[g2]> it defaults to the openslug ipkgs feed
Jun 30 12:10:33 <[g2]> that's my understanding anyway
Jun 30 12:10:40 <[g2]> I always build and run my own
Jun 30 12:10:53 * beewoolie-afk sheepishly turns the HD on.
Jun 30 12:11:01 <[g2]> :)
Jun 30 12:11:09 <beewoolie-afk> I would build, but I don't have the time right now.
Jun 30 12:11:29 <beewoolie-afk> I'm just hoping to get it online today so I can prepare for a baseline performance measurement.
Jun 30 12:12:22 <beewoolie-afk> Also, I'd be really appreciative if you'd add the IPv6 modules to the kernel build.
Jun 30 12:13:53 <[g2]> I think someone's been playing with ipv6
Jun 30 12:13:59 <[g2]> I haven't played at all
Jun 30 12:14:14 <[g2]> I don't have any IPv6 setup
Jun 30 12:14:23 <beewoolie-afk> It's really sweet.  I can ssh into my notebook through the IPv4 NAT gateway.
Jun 30 12:15:37 <[g2]> dynamically ?
Jun 30 12:16:06 <[g2]> you can do the same thing with port routing and dhcp on IPv4
Jun 30 12:16:16 <[g2]> port forwarding
Jun 30 12:16:25 <[g2]> or static for that matter
Jun 30 12:16:32 <beewoolie-afk> yeah, but this is direct addressing.
Jun 30 12:16:48 <beewoolie-afk> No NAT.  Anyone who plugs into my network gets the same access.
Jun 30 12:17:05 <beewoolie-afk> I've been chatting on a couple of ipv6 lists about the dynamic DNS and IPSEC issues.
Jun 30 12:17:16 <beewoolie-afk> Turns out that not all of the groundwork is complete.
Jun 30 12:17:17 <[g2]> it's IPv6-IPv4-IPv6 gatewaying
Jun 30 12:17:34 <beewoolie-afk> There are a couple of methods.  i use a hybrid.
Jun 30 12:17:47 <[g2]> you're tunneling IPv6 over and IPv4 backbone
Jun 30 12:17:48 <beewoolie-afk> http://wiki.buici.com/twiki/bin/view/Main/PracticalIPv6Guide
Jun 30 12:18:35 <beewoolie-afk> [g2]: ipkg search samba returns nothing.
Jun 30 12:18:48 <[g2]> ipkg list | grep samba
Jun 30 12:19:00 <[g2]> and you did you ipkg update ?
Jun 30 12:19:05 <beewoolie-afk> yeah.
Jun 30 12:19:05 <[g2]> your ipkg
Jun 30 12:19:13 <beewoolie-afk> Hmm.  Search doesn't work, but list does.
Jun 30 12:19:41 <[g2]> I don't think search is a command for ipkge
Jun 30 12:19:44 <beewoolie-afk> Very exciting!
Jun 30 12:19:54 <[g2]> welcome aboard :)
Jun 30 12:20:04 <[g2]> are you running a 2.0 ?
Jun 30 12:20:14 <beewoolie-afk> I've been wanting to get back to this for some time.  I need the storage and a lack of access has been making it hard.
Jun 30 12:20:19 <beewoolie-afk> How do I know which version I have?
Jun 30 12:20:48 <[g2]> it prints out when you boot and I think there's something in /etc
Jun 30 12:22:06 <[g2]> cat /etc/issue
Jun 30 12:22:08 <beewoolie-afk> I'm getting ldconfig errors.
Jun 30 12:22:19 <beewoolie-afk> It isn't in the path!
Jun 30 12:22:31 <[g2]> the samba in the feeds may be broke
Jun 30 12:22:45 <beewoolie-afk> I think it's a setup issue.  /sbin isn't in the root path.
Jun 30 12:22:48 <[g2]> nail setup and checked in the changes in monotone
Jun 30 12:23:04 <[g2]> right, the tweaked the paths
Jun 30 12:23:11 <beewoolie-afk> OpenSlug 1.12-beta 
Jun 30 12:23:15 <[g2]> I think it's all fixed in monotone
Jun 30 12:23:17 <[g2]> that's old
Jun 30 12:23:24 <beewoolie-afk> Can I upgrade, or do I have to reinstall?
Jun 30 12:23:33 <[g2]> that need re-installed
Jun 30 12:23:38 <beewoolie-afk> Hmm.  
Jun 30 12:23:46 <[g2]> but once you are at 2.0 you can upgrade from OpenSlug
Jun 30 12:23:59 <beewoolie-afk> I guess this isn't going to happen today.
Jun 30 12:25:29 <beewoolie-afk> bbl
Jun 30 12:26:56 <[g2]> VoodooZ_Work, what command did you used for measuring the nc
Jun 30 12:27:12 <[g2]> time cat foo | nc ....
Jun 30 12:30:35 <VoodooZ_Work> time cat file | nc address port on one side
Jun 30 12:30:58 <VoodooZ_Work> and nc -l port > /dev/null on the receiving side.
Jun 30 12:31:24 <VoodooZ_Work> But I noticed a difference in sytanx between fedora core4's nc and that one in the slug ...
Jun 30 12:31:29 <[g2]> I'm writing to disk on the target
Jun 30 12:31:38 <jacques> anyone here familiar with the wiley hardware ?
Jun 30 12:31:51 <[g2]> not I said [g2] 
Jun 30 12:31:53 <VoodooZ_Work> I wanted to isolate network speed so..
Jun 30 12:31:55 <jacques> I think I read that wl-500g (non-dlx) was buggy hardware and I'm trying to confirm
Jun 30 12:32:12 <[g2]> I'm gonna build ttcp
Jun 30 12:32:43 <[g2]> however, I'm on a new switch
Jun 30 12:32:48 <VoodooZ_Work> nc can be used for quite a few things.
Jun 30 12:32:53 <[g2]> I don't know how fast that is
Jun 30 12:33:01 <[g2]> well the real nc
Jun 30 12:34:37 <VoodooZ_Work> what?
Jun 30 12:35:03 <VoodooZ_Work> I just divided the file size in bytes by the time in secs.
Jun 30 12:36:52 <VoodooZ_Work> anyways, I have to leave work now as I'm going camping for Canada's Day tommorow so I'll see you all on Sunday.
Jun 30 12:37:00 <VoodooZ_Work> Later...
Jun 30 12:44:59 <[g2]> hmmm interesting
Jun 30 12:45:09 <[g2]> on TurboFatty:
Jun 30 12:45:27 <[g2]> ./hdparm -Tt /dev/sda
Jun 30 12:45:28 <[g2]> /dev/sda:
Jun 30 12:45:28 <[g2]>  Timing cached reads:   216 MB in  2.01 seconds = 107.46 MB/sec
Jun 30 12:45:28 <[g2]>  Timing buffered disk reads:   38 MB in  3.32 seconds =  11.45 MB/sec
Jun 30 12:46:17 <jacques> ah
Jun 30 12:46:29 <jacques> yeah you get more beause it likes to do bigger reads than I have ram
Jun 30 12:46:35 <jacques> I can see it stutter at the end
Jun 30 12:46:52 <[g2]> I've got a Maxtor External
Jun 30 12:46:57 <jacques> so I only get 10MB/s
Jun 30 12:47:01 <[g2]> I'm sure different disks matter
Jun 30 12:47:12 <jacques> not that much
Jun 30 12:47:22 <[g2]> OH the cornice is SLOW
Jun 30 12:47:28 <jacques> try that on a non-fat slug and get back to me
Jun 30 12:47:37 <jacques> I bet it's more like 10MB/s
Jun 30 12:47:44 <[g2]> probably
Jun 30 12:47:55 <[g2]> the interesting thing was TCP throughput
Jun 30 12:48:26 <[g2]> ttcp to /dev/null was ttcp-t: 223150080 bytes in 19.03 real seconds = 11.18 MB/sec +++
Jun 30 12:48:26 <[g2]> ttcp-t: 27240 I/O calls, msec/call = 0.72, calls/sec = 1431.68
Jun 30 12:48:35 <[g2]> but do disk was 
Jun 30 12:48:40 <jacques> well I just got this :
Jun 30 12:48:42 <jacques> /dev/sda:
Jun 30 12:48:42 <jacques>  Timing cached reads:   216 MB in  2.00 seconds = 108.00 MB/sec
Jun 30 12:48:42 <jacques>  Timing buffered disk reads:   34 MB in  3.11 seconds =  10.93 MB/sec
Jun 30 12:48:42 <jacques> r
Jun 30 12:48:47 <[g2]> ttcp-t: 223150080 bytes in 44.06 real seconds = 4.83 MB/sec +++
Jun 30 12:48:47 <[g2]> ttcp-t: 27240 I/O calls, msec/call = 1.66, calls/sec = 618.19
Jun 30 12:49:09 <jacques> 11.18 MB/sec wow that's gotta be max
Jun 30 12:49:28 <[g2]> I'd guess .5MB or 1/22 is just disk differences
Jun 30 12:49:47 <jacques> well at the end of the disk read, can you see your disk light stutter? I can
Jun 30 12:50:01 <jacques> if it was reading like 20MB I think my number would be faster
Jun 30 12:50:04 <jacques> higher
Jun 30 12:50:31 <[g2]> well from the PC the disk is 60MBs
Jun 30 12:50:58 <jacques> through usb or ide ?
Jun 30 12:51:07 <[g2]> that's the native disk
Jun 30 12:51:18 <[g2]> I think I saw close to 20 with the USB
Jun 30 12:51:30 <[g2]> I'll test it on the PC
Jun 30 12:52:26 <jacques> that sounds right - same as I get
Jun 30 12:52:36 <jacques> about 22MB/s usb, 57MB/s ide
Jun 30 12:53:47 <[g2]> hdparm -tT /dev/sdb
Jun 30 12:53:47 <[g2]> /dev/sdb:
Jun 30 12:53:47 <[g2]>  Timing cached reads:   3340 MB in  2.00 seconds = 1669.42 MB/sec
Jun 30 12:53:47 <[g2]>  Timing buffered disk reads:   88 MB in  3.05 seconds =  28.83 MB/sec
Jun 30 12:54:05 <[g2]> that from the USB port
Jun 30 12:54:08 <jacques> nice
Jun 30 12:54:42 <[g2]> it's really no biggie....
Jun 30 12:55:10 <[g2]> the point to me is I'm only seeing 5MB via ttcp to the disk
Jun 30 12:55:27 <[g2]> I wonder if running in a poll mode would provide much more through pot
Jun 30 12:55:29 <[g2]> :)
Jun 30 12:55:33 <[g2]> throughput
Jun 30 12:55:56 <jacques> easiest way to find out is probably to try it
Jun 30 12:56:19 <[g2]> the other question is verifying write speeds
Jun 30 12:57:09 <[g2]> time dd if=/dev/zero of=file bs=512 count=200k
Jun 30 12:58:14 <[g2]> WHOA
Jun 30 12:58:26 * [g2] thought that was too good to be true ;)
Jun 30 12:58:37 <jacques> need to sync
Jun 30 12:58:39 <jacques> heh
Jun 30 12:58:42 <[g2]> time dd if=/dev/zero of=file bs=1M count=128
Jun 30 12:58:42 <[g2]> 128+0 records in
Jun 30 12:58:42 <[g2]> 128+0 records out
Jun 30 12:58:42 <[g2]> real    0m0.342s
Jun 30 12:59:02 <[g2]> wrong computer :) on the PC
Jun 30 12:59:34 <jacques> still, I doubt even your pc can write 128MB in 0.3 seconds
Jun 30 12:59:42 <jacques> it just went into ram
Jun 30 12:59:58 <[g2]> nod
Jun 30 13:01:27 <[g2]> time dd if=/dev/zero of=file bs=1M count=128; sync
Jun 30 13:01:27 <[g2]> 128+0 records in
Jun 30 13:01:27 <[g2]> 128+0 records out
Jun 30 13:01:27 <[g2]> real    0m 9.97s
Jun 30 13:01:43 <[g2]> which is about 12MB/sec
Jun 30 13:04:00 <[g2]> ok... 
Jun 30 13:04:17 <[g2]> that writing takes up like 75-85% of the CPU
Jun 30 13:04:48 <[g2]> so basically, the eth0 interrupt handler and the disk are contending for CPU
Jun 30 13:05:09 <[g2]> I read and interesting little tidbit in the dev manual
Jun 30 13:05:28 <[g2]> the mentioned locking the stack of an interrupt handler in the mini data cache
Jun 30 13:05:36 <[g2]> which runs a core speed
Jun 30 13:05:39 <[g2]> at
Jun 30 13:05:56 <jacques> neat
Jun 30 13:06:18 <[g2]> I'd think the eth0 or possibly the usb hander might be candidates for that
Jun 30 13:08:35 <jacques> that sounds really cool
Jun 30 13:09:12 <ep1220> g2: can You tell me base address and size of the RAM on the slug ?
Jun 30 13:13:09 <jbowler-zzz> ep1220: cached or uncached?
Jun 30 13:13:18 <ep1220> uncached
Jun 30 13:14:09 <ep1220> is 32MB starting at 0x100000 correct ?
Jun 30 13:15:06 <jacques> hi jbowler-zzz you've been kinda quiet today
Jun 30 13:15:41 <jbowler-zzz> It's table 106 in the dev manual.
Jun 30 13:16:12 <jbowler-zzz> Sorry, wrong table...  it's in here somewhere ;-)
Jun 30 13:17:54 <jbowler-zzz> I'm pretty sure that address is wrong.  I think it may be 0x20000000
Jun 30 13:18:35 <ep1220> I already put some program at 0x100000 - so i think there is RAM there.
Jun 30 13:18:50 <ep1220> that table you pointed me too, might be the right one.
Jun 30 13:20:41 <jbowler-zzz> There's ram everything in the first 1GByte, but only 256MByte of it is uncached iirc.
Jun 30 13:20:47 <jbowler-zzz> s/everything/everywhere/
Jun 30 13:24:28 <jbowler-zzz> I think I remember now, it's determined by the boot loader because the page table determines the caching (iirc).
Jun 30 13:24:56 <ep1220> Likely.
Jun 30 13:25:20 <ep1220> maybe non-cached memory is mapped in the PCI area only (?)
Jun 30 13:32:11 <jbowler-zzz> Here's what RedBoot does in the stock version (I don't think this is the LinkSys hacked version, but that's probably the same):
Jun 30 13:32:12 <jbowler-zzz>         IXP_MAP_SDRAM   SDRAM_BASE,           1, 0, 0, 0   // Cached SDRAM
Jun 30 13:32:12 <jbowler-zzz>         IXP_MAP_SDRAM   SDRAM_ALIAS_BASE,     1, 0, 0, 0   // Cached SDRAM alias
Jun 30 13:32:12 <jbowler-zzz>         IXP_MAP_SDRAM   SDRAM_UNCACHED_BASE,  0, 0, 0, 0   // Uncached SDRAM
Jun 30 13:32:12 <jbowler-zzz>         IXP_MAP_SDRAM   SDRAM_DC_BASE,        1, 0, 0, 1   // Cached data coherent SDRAM
Jun 30 13:33:07 <ep1220> thx
Jun 30 13:33:28 <jbowler-zzz> That's the setup for the four physical replications of the 256MByte SDRAM space - so addresses 0, 0x10000000, 0x20000000, 0x30000000
Jun 30 13:34:05 <jbowler-zzz> iirc some of the 256MByte address lines aren't connected on the NSLU2, so that causes replication within the 256MByte.
Jun 30 13:34:37 <ep1220> there will be a 32MEG area, replicated every 256MB
Jun 30 13:34:43 <jbowler-zzz> hi jacques: busy working, just came in for lunch!
Jun 30 13:35:07 <jacques> jbowler-zzz, I guess you gotta work sometime :-)
Jun 30 13:35:29 <jacques> jbowler-zzz, you gonna turbo a slug?
Jun 30 13:35:38 <jbowler-zzz> ep1220: yes, plus within the 256MByte there will be replications (I think).
Jun 30 13:36:32 <jbowler-zzz> jacques: I'm going to swap R64 to R81, when I have time - that'll give the 'official' setting b100 from the app note.  I'm pretty much 99.99% certain the result will be the same.
Jun 30 13:37:00 <jacques> jbowler-zzz, cool, more data
Jun 30 13:37:22 <jbowler-zzz> back to work... bbl
Jun 30 13:37:25 <Tiersten> So technically removing R83 isn't what you're supposed to do?
Jun 30 13:37:48 <jbowler-zzz> well, you're not supposed to open the case...
Jun 30 13:37:53 <Tiersten> heh
Jun 30 13:37:57 <jacques> lol I was just gonna say somethin glike that
Jun 30 13:38:18 <jbowler-zzz> b011 is a documented setting, in the dev manual, but the boot time app note doesn't mention it - just b100.
Jun 30 13:38:19 <Tiersten> You mean my warranty is void now that I've added JTAG, serial, 2nd serial and drilled a hole in the back for a serial cable?  Dang
Jun 30 13:38:25 <Tiersten> ah
Jun 30 13:47:37 <jacques> I have lmbench data comparing stock and full-speed slugs - where shall I put it on the wiki ?
Jun 30 13:49:50 <jacques> Tiersten, what's the proper way to do preformatted text on the wiki again?
Jun 30 13:50:24 <Tiersten> <space>[= stuff =]
Jun 30 13:50:30 <Tiersten> That *should* do most stuff correctly
Jun 30 13:51:00 <jacques> Tiersten, but I have about a page of text - lmbench results
Jun 30 13:51:22 <jacques> does what you said work for multi-line?
Jun 30 13:51:28 <Tiersten> It should do...
Jun 30 13:51:37 <Tiersten> If not then you'll have to do it each line I'm afraid
Jun 30 13:51:44 <jacques> sigh
Jun 30 13:51:48 <jacques> this has to be a bug
Jun 30 13:51:57 <Tiersten> ?
Jun 30 13:52:33 <jacques> there has to be a better way
Jun 30 13:52:38 <jacques> is what I mean
Jun 30 13:52:53 <Tiersten> Upload it as a file and then link to it
Jun 30 13:53:45 <jacques> wow it appears to be working
Jun 30 13:55:29 <jacques> http://www.nslu2-linux.org/wiki/HowTo/OverClockTheSlug at the bttom
Jun 30 13:59:52 <[g2]> jacques is the MAN!
Jun 30 13:59:56 <[g2]> nice stats!
Jun 30 14:08:22 <jacques> i ran them a couple of times to make sure they aren't toally bogus
Jun 30 14:09:04 <jacques> tho they are using slightly different kernel versions and glibc versions  :-\
Jun 30 14:13:18 <beewoolie-afk> jbowler-zzz: I considered what you suggest, moving r64 to R81.  IMHO, there isn't anything to be gained.
Jun 30 14:13:40 <beewoolie-afk> I think that b000 is documented, tho.
Jun 30 14:14:05 <beewoolie-afk> Nice stats, too.
Jun 30 14:18:47 <[g2]> jacques, ping
Jun 30 14:19:30 <[g2]> when you get a chance type "top"
Jun 30 14:20:26 <jacques> [g2], hmm what do you need to know?  I have procps top installed
Jun 30 14:20:43 <[g2]> bb top says:
Jun 30 14:20:46 <[g2]> Load average: 0.99, 0.71, 0.32    (State: S=sleeping R=running, W=waiting)
Jun 30 14:20:54 <[g2]> the numbers are < 1
Jun 30 14:21:00 <jacques> hmmmmmmmm
Jun 30 14:21:12 <[g2]> I thought you'd get a kick out of that
Jun 30 14:21:22 <jacques>  09:24:22 up 11:50, load average: 1.00, 1.02, 1.14
Jun 30 14:21:26 <jacques> :-(
Jun 30 14:21:35 <jacques> did you change something?
Jun 30 14:21:40 <[g2]> nope
Jun 30 14:21:58 <[g2]> could be my time was out of sync before
Jun 30 14:22:07 <[g2]> but I'd don't imagine that's it
Jun 30 14:57:40 <rwhitby-asleep> beewoolie-afk: the reason why the official mod is R83, is because that sets the config to 011 which is a documented value.  R64 caused a working, but undocumented in the data book, value.
Jun 30 14:58:08 <rwhitby> (that's assuming these guys gave the correct resistor or config bit mapping ...)
Jun 30 14:58:55 <dyoung-zzzz> beewoolie-afk, sorry I gave bad info yesterday, turns out R83 should be what I reported as R81
Jun 30 14:59:25 <rwhitby> R83 is pin 3 on the flash
Jun 30 14:59:49 <dyoung-zzzz> Also I have thermal reports.  After like at least 8 hours sitting  under a CD Jewel Case the max temperature on the special spot 5mm to the right of the I was 53degC
Jun 30 15:00:00 <dyoung-zzzz> thats with a loadavg of 2.
Jun 30 15:00:46 <jacques> yeah I'm really starting to think heat will not be an issue
Jun 30 15:00:48 <rwhitby> and my three TurboSlugs have been running all night, with no discernable increase in temperature on the outside of the case with my extremely accurate "feel it with the fingers" temperature probe.
Jun 30 15:01:02 <jacques> same here even when compiling for an hour
Jun 30 15:01:14 <rwhitby> heh, I thought it from the start.  how hot can a 266MHz processor on 0.13micron get ...
Jun 30 15:01:25 <jacques> true
Jun 30 15:01:30 <Tiersten> I've pointed my IR thermometer at mine and it's not exceeded 50C
Jun 30 15:01:57 <dyoung-zzzz> I'm satisifed that its not going to meltdown, so I'm gonna do the others now.
Jun 30 15:01:59 <jacques> like I said, I've seen my pentium M get over 80C under load
Jun 30 15:02:09 <dyoung-zzzz> what do I have a chance for, #15?
Jun 30 15:02:18 <rwhitby> 50C is nothing.
Jun 30 15:02:26 <jacques> I agree
Jun 30 15:02:53 <rwhitby> we need a link to the database on the TurboSlug wiki page
Jun 30 15:03:22 <jacques> my athlon XPs in my dual were also running > 80C under cpuburn before I got the nice heatsinks and arctic silver paste
Jun 30 15:03:38 <jacques> and AMD sait they are good to 85C
Jun 30 15:04:07 <rwhitby> so no-one outside "the dev team" has turboed yet.
Jun 30 15:04:11 <jacques> (this was measured at the internal chip sensor, not the socket thermistor)
Jun 30 15:05:03 <jacques> I guess heat isn't an issue because none of the intel arm chips have internal temperature sensors :-D
Jun 30 15:05:12 <rwhitby> dyoung-zzzz: dvd finished
Jun 30 15:05:32 <dyoung-zzzz> Cool.
Jun 30 15:05:38 <dyoung-zzzz> Shall I send you some blank ones?
Jun 30 15:05:44 <jacques> rwhitby, did you have one of your turboslugs transcode a dvd ??
Jun 30 15:07:25 <jacques> rwhitby and dyoung-zzzz take a look at lmbench results at bottom of http://www.nslu2-linux.org/wiki/HowTo/OverClockTheSlug
Jun 30 15:07:59 <rwhitby> jacques: no, just serve up the files and take the output to disk
Jun 30 15:08:28 <jacques> ah
Jun 30 15:09:14 <rwhitby> that's what it was doing before it went into surgery, so I had to restart it
Jun 30 15:10:01 <Tiersten> What is the difference between removing R64 and R83?  The bits are set differently?
Jun 30 15:12:55 <dyoung-zzzz> nice lm bench numbers
Jun 30 15:13:23 <dyoung-zzzz> tiersten, yes.  Just one bitset is listed as supported in the DS, the other not (but it works anywayss)
Jun 30 15:13:47 <Tiersten> Ah okay.  They're equivalent to the same speed?
Jun 30 15:14:20 <dyoung-zzzz> as far as I can tell, yes.
Jun 30 15:15:13 <Tiersten> k
Jun 30 15:16:02 <rwhitby> R83 is less dangerous physically - no tracks under or parallel to it (unlike R64)
Jun 30 15:16:27 <Tiersten> I'm not concerned about that.  Just wondering about the difference :)
Jun 30 15:16:56 <rwhitby> documented value was the main reason.  physical safety confirmed the decision
Jun 30 15:17:24 <dyoung-zzzz> rwhitby was concerned about ph ysical safety because he was using a 1cm tip iron. :-)
Jun 30 15:17:41 <Tiersten> Is there any point in trying > 266MHz?  I know rwhitby said it was locked from the factory
Jun 30 15:17:50 <Tiersten> yeah.  with no temperature adjustment either :)
Jun 30 15:18:16 <rwhitby> can try, but it won't work.  fuses on-chip according to the databook
Jun 30 15:18:38 <jacques> you can apply 12V to certain pins to burn out the fuses tho
Jun 30 15:19:19 <rwhitby> I doubt it.  they are usually laser fuses done before packaging
Jun 30 15:19:57 <dyoung-zzzz> someone was talking about overclocking earlier.
Jun 30 15:20:24 <Tiersten> Just wondering if it's actually locked for certain or whether the book just says that to discourage you
Jun 30 15:20:27 <Tiersten> yeah.  voodooz
Jun 30 15:20:36 <Tiersten> He wanted to replace the 33MHz oscillator on it with a 40MHz
Jun 30 15:20:44 <dyoung-zzzz> with a little bit of work, one hsould be able to clock the pci externally at 33Mhz and change the master clock to something else.
Jun 30 15:20:45 <Tiersten> And not have all the buses go crazy
Jun 30 15:20:55 <dyoung-zzzz> Unless the PCI will actually work at 40Mhz
Jun 30 15:21:06 <dyoung-zzzz> (and everything else for that matter)
Jun 30 15:21:25 <Tiersten> Doubt it.  jacques has pointed out that the NEC USB chip is flakey at anything other than 33MHz
Jun 30 15:21:58 <rwhitby> here's a conspiracy theory: Linksys deliberately underclocked it so the product sales would benefit from mid-life publicity about doubling the speed.  and we are just pawns in a nefarious marketing plan.
Jun 30 15:22:12 <Tiersten> Yeah.  That's what I said earlier :)
Jun 30 15:22:39 <dyoung-zzzz> so if GPIO (forget which one) is available, you can feed it the 33Mhz external PCI Clock; then replace this 33Mhz master dclock with a 40Mhz and have a hope of it working.
Jun 30 15:22:40 <Tiersten> NSLU2.5 and it'll be just the same PCB with the resistor moved  and a newer case :)
Jun 30 15:22:56 <rwhitby> USB 2.5?
Jun 30 15:23:03 <dyoung-zzzz> and using the new stock of NEC Host Controller that doesnt suck?
Jun 30 15:23:17 <Tiersten> I think all NEC controllers suck...
Jun 30 15:23:27 <Tiersten> GPIO14 is the PCI clock
Jun 30 15:23:47 <jacques> actually I was looking at the board image, and the NEC chip ends in 101 and I think that might have been the "better" one :-(  the old one ending in 100
Jun 30 15:23:51 <Tiersten> How are you going to clock the IXP internal PCI circuitry?
Jun 30 15:24:16 <Tiersten> Since just clocking the NEC at 33MHz whilst the IXP is internally 40MHz won't do very much
Jun 30 15:24:34 <Tiersten> Actually, is there a PCI clock input on the IXP?
Jun 30 15:24:35 <dyoung-zzzz> Theres a setting to take a external signal to use as the PCI Clock.  Its unclear to me what exactly thats doing
Jun 30 15:24:43 <dyoung-zzzz> yeah, PCI Input Clock.
Jun 30 15:24:49 <dyoung-zzzz> its another boot setting.
Jun 30 15:24:53 <Tiersten> The "external" signal is actually the GPIO
Jun 30 15:25:02 <Tiersten> 14 and 15 are special as they can be configured to be oscillators
Jun 30 15:25:08 <dyoung-zzzz> Yes, thats what I mention "if its available"
Jun 30 15:25:29 <Tiersten> I don't think it actually goes out from under the IXP
Jun 30 15:25:36 <Tiersten> I keep forgetting to update the wiki page about that
Jun 30 15:28:14 <dyoung-zzzz> Ah well.  Was a passing thought.
Jun 30 15:30:35 <jacques> http://www.necel.com/usb/en/product/upd720101.html yeah I think ours actually is the later rev
Jun 30 15:33:08 <Tiersten> ahh okay
Jun 30 15:33:42 <jacques> the 100A was total crap
Jun 30 15:34:20 <dyoung-zzzz> I wonder if later reviions suck less th ough, ours is D720101 F1, I wonder if theres a G or H or something that works better
Jun 30 15:34:43 <Tiersten> There is a D720101GJ on my terastation
Jun 30 15:35:08 <Tiersten> It's not exactly a speed demon on there either but it's acceptable
Jun 30 15:35:15 <dyoung-zzzz> heh
Jun 30 15:35:36 <dyoung-zzzz> so linksys probably bought all the old stuff for cheap so they can sell these things at a $80 price point
Jun 30 15:35:46 <Tiersten> bastards :)
Jun 30 15:35:56 <dyoung-zzzz> so the nslu2 Rev 2 will have a faster core and a USB chip that works.
Jun 30 15:36:11 <jacques> and 64MB RAM
Jun 30 15:36:21 <dyoung-zzzz> Right.
Jun 30 15:36:51 <Tiersten> The extra RAM might be a possibility.  Buffalo keep adding RAM to their Linkstations as they come out with newer versions
Jun 30 15:36:58 <jacques> any of you know if the non-dlx wl-500g should be avoided - I think I read it was buggy hardware
Jun 30 15:38:18 <rwhitby> mine worked fine while I had it. 
Jun 30 15:38:34 <rwhitby> but I have a 10MB cable modem interface anyeay
Jun 30 15:47:02 <beewoolie-afk> dyoung-zzzz: what makes you think there's going to be a rev2?
Jun 30 15:47:42 <rwhitby> there will never be a rev2.
Jun 30 15:49:09 <Tiersten> We don't think so.  It's just that it may just be a sinister plot to release an "upgraded" NSLU2 which just has the resistor removed
Jun 30 15:49:20 <beewoolie-afk> rwhitby: If I may, I think the caution about tracks under r64 maybe overly conservative.
Jun 30 15:50:45 <rwhitby> beewoolie-afk: for you and I yes.  for Joe User who uses a sharp knife to remove the resistor, no :-)
Jun 30 15:51:08 <Tiersten> Sharp knife?  I'm more thinking big kitchen knife :)
Jun 30 15:51:13 <beewoolie-afk> Um, sharp knife?  Perhaps a pair of diagonal cutters?
Jun 30 15:51:30 <beewoolie-afk> I find it easier to desolver with room around the resistor.
Jun 30 15:51:38 <beewoolie-afk> s/desolver/desolder/
Jun 30 15:52:05 <Tiersten> If you were especially insane you could probably do the mod with a dremel
Jun 30 15:52:21 <ByronT> except when you flick the resistor onto the floor!
Jun 30 15:52:27 <Tiersten> :)
Jun 30 15:52:28 <beewoolie-afk> May I suggest that we make it clear that removing r83 or r64 will work.  R83 being the less risky choice for the novice.
Jun 30 15:52:42 <ByronT> and then try to locate it for 15 minutes on hands and knees
Jun 30 15:52:50 <rwhitby> beewoolie-afk: why suggest R64 when the combination that produces is not listed in the databook?
Jun 30 15:53:02 <beewoolie-afk> I thought that 000 was listed?
Jun 30 15:53:15 <jbowler-zzz> The resistors are 0's, not 1's
Jun 30 15:53:35 <jbowler-zzz> So -r64 -> 110
Jun 30 15:53:52 <beewoolie-afk> Ah, that wasn't clear.  
Jun 30 15:54:40 <beewoolie-afk> Right.  I should have known.  I think I chose R64 because I figured that if I had to make it b100, then it was better to start with removing the unneeded resistor before placing it at R82.
Jun 30 15:54:45 <beewoolie-afk> I got lucky and it works.
Jun 30 15:55:05 <beewoolie-afk> So, removing R83 make it b011.
Jun 30 15:55:26 <rwhitby> R64 certainly does work, but it's a bit harder to explain on the wiki _why_ R64 works :-)
Jun 30 15:55:27 <beewoolie-afk> I still think it is worth documenting what we've discovered, even if it may not work for everyone.
Jun 30 15:55:32 <rwhitby> nod
Jun 30 15:55:59 <beewoolie-afk> Certainly, there is a chance that some rev of the CPU won't work with the multiplied at b110.
Jun 30 15:56:11 <rwhitby> with the caveat that Intel could change the behaviour of an R64 removal at any time with a new minor rev of the chip
Jun 30 15:56:45 <beewoolie-afk> That's what I just wrote.
Jun 30 15:57:33 <rwhitby> yeah, snap :-)
Jun 30 15:58:33 <dyoung-zzzz> we should probably add to the wiki "Why does this work".  I said I would do it, but I woke up really late.
Jun 30 15:59:24 <beewoolie-afk> There isn't really a hurry.
Jun 30 16:01:57 <jacques> so the underclocked setting is documented somewhere ?
Jun 30 16:02:15 <ByronT> yeah
Jun 30 16:02:26 <ByronT> let me dig up the URL to the doco
Jun 30 16:03:13 <ByronT> http://www.intel.com/design/network/manuals/252480.htm page 324-326 I believe
Jun 30 16:03:39 <jacques> thanks ByronT 
Jun 30 16:04:52 * cbpage dlurks to say that he had a board at work R83-modded just now... worked right away - thanks!
Jun 30 16:05:46 <jacques> OK looking at page 326, all the settings say 266MHz core speed
Jun 30 16:06:04 <beewoolie-afk> The one used by Linksys b010 isn't documented.
Jun 30 16:06:13 <rwhitby> look at the 533 to see the divide by 2 setting which Linksys used
Jun 30 16:06:34 <rwhitby> glc just used a fingernail clipper to remove R83 :-)
Jun 30 16:07:01 <jacques> right so we're still reading between the lines - underclocking of the 266MHz part isn't documented anywhere that we've found
Jun 30 16:07:06 <beewoolie-afk> Oh my.  We've created some monsters.
Jun 30 16:07:15 <Tiersten> Monsters?
Jun 30 16:07:21 <rwhitby> beewoolie-afk: *that's* why it's R83 :-)
Jun 30 16:07:33 <Tiersten> give me a minute and I'll document it
Jun 30 16:07:54 <beewoolie-afk> Somehow, I'm insulted by the barbaric treatment of the lowly slugs.
Jun 30 16:08:03 <jacques> the point I was working towards is, isn't it more likely that the undocumented underclocking feature will cease to work in future revs?
Jun 30 16:08:49 <rwhitby> no, cause it is documented for the 533 binned parts
Jun 30 16:09:10 <jacques> you are using a strange defiition of documented
Jun 30 16:09:12 <Tiersten> It's not documented for the 266MHz binned parts but chances of it breaking are very unlikely
Jun 30 16:09:18 <beewoolie-afk> My copy of the manual doesn't show 0b010.
Jun 30 16:09:21 <beewoolie-afk> Even for 533.
Jun 30 16:09:24 <jacques> I am looking at a table, nowhere does it show a 133MHz core clock
Jun 30 16:09:47 <rwhitby> Tiersten: it will be the same die
Jun 30 16:09:54 <beewoolie-afk> This is the march 2005 version.
Jun 30 16:09:55 <Tiersten> I know.  That's what I was trying to say :)
Jun 30 16:09:57 <dyoung-zzzz> jacques, the linksys configuration is undocumneted.
Jun 30 16:10:20 <beewoolie-afk> rwhitby said that   look at the 533 to see the divide by 2 setting which Linksys used
Jun 30 16:10:20 <dyoung-zzzz> so a small amount of guessing is involved.
Jun 30 16:10:35 * rwhitby lied :-)
Jun 30 16:10:46 <beewoolie-afk> OK.  Just making sure we're on the same page.
Jun 30 16:10:51 <beewoolie-afk> Literally.
Jun 30 16:10:57 <Tiersten> R84 = A23.  What are the others?
Jun 30 16:11:05 <beewoolie-afk> R64 == A23
Jun 30 16:11:17 <Tiersten> I thought that was wrong?
Jun 30 16:11:22 <beewoolie-afk> Oh, really?
Jun 30 16:11:25 <Tiersten> Yeah
Jun 30 16:11:27 <beewoolie-afk> dyoung-zzzz: ?
Jun 30 16:11:32 <Tiersten> R64 = A21 or somethig
Jun 30 16:11:41 <beewoolie-afk> You mean that removing R64 is way out of control?
Jun 30 16:11:46 * Tiersten *shrugs*
Jun 30 16:11:47 <ByronT> glc with nail clippers?
Jun 30 16:11:52 <ByronT> ha!
Jun 30 16:11:54 <beewoolie-afk> I got lucky?
Jun 30 16:12:02 <rwhitby> heh, you'd think with 13 turboslugs we'd have a better idea of which resistor is which :-)
Jun 30 16:12:03 <Tiersten> Okay.  Lets start over
Jun 30 16:12:07 <beewoolie-afk> eeeny meenie miney moe...
Jun 30 16:12:08 <dyoung-zzzz> no the error was in my intepretation of R83/R82.
Jun 30 16:12:14 <Tiersten> Ah okay
Jun 30 16:12:25 <beewoolie-afk> This is something I'd really like to see on the wiki.
Jun 30 16:12:27 <Tiersten> Okay.  Can you tell me which resistors are what :)
Jun 30 16:12:32 <dyoung-zzzz> Okay.
Jun 30 16:12:32 <Tiersten> beewoolie-afk: This is what I'm trying to document
Jun 30 16:12:40 <beewoolie-afk> ;-)
Jun 30 16:12:50 <ByronT> beewoolie-afk, we're guessing that you got lucky - I'm guessing that the core reads your setting as invalid and simply goes to default speed
Jun 30 16:13:19 <beewoolie-afk> Well, it was on the way to being 0b100.
Jun 30 16:13:30 <beewoolie-afk> My next step would have made it compliant.
Jun 30 16:13:44 <rwhitby> nod
Jun 30 16:13:57 <Tiersten> soooo.... umm... which resistor = what? :)
Jun 30 16:14:11 <ByronT> remove r83 is all I know... ;)
Jun 30 16:14:13 <rwhitby> R83 is flash chip pin 3
Jun 30 16:14:14 <beewoolie-afk> Take it away dyoung-zzzz ...
Jun 30 16:14:15 <Tiersten> I looked through the logs and it all seems to conflict
Jun 30 16:14:50 <beewoolie-afk> There is a strong indication that R83 is A21, R64 is A23, and R82 is A22.
Jun 30 16:15:08 <beewoolie-afk> I believe that the R83mod'd slugs report 0b011 as the multiplier.
Jun 30 16:15:08 <ByronT> well, dyoung's typing errors made the logs look funny
Jun 30 16:15:27 <beewoolie-afk> dyoung-zzzz: I've got your back, man.
Jun 30 16:15:31 <ByronT> I can't tell you the number of times, I went back and consulted the information and then the corrections....
Jun 30 16:15:50 <Tiersten> And the default configuration by Linksys is 011?
Jun 30 16:15:58 <dyoung-zzzz> A23 = R64 A22 = R81 A21 = R83
Jun 30 16:16:04 <beewoolie-afk> No, the default config by linksys is 0b010.
Jun 30 16:16:24 <beewoolie-afk> [g2] verified that when he read the register.  0x004...
Jun 30 16:16:31 <Tiersten> Ah okay
Jun 30 16:16:32 <beewoolie-afk> The avila reports 0x008....
Jun 30 16:16:38 <Tiersten> R81 being the unpopulated spot
Jun 30 16:16:38 <beewoolie-afk> Which is 0b100
Jun 30 16:16:48 <beewoolie-afk> Right.  I might have that number wrong.
Jun 30 16:16:54 <beewoolie-afk> It's R83 and R81 and R64.
Jun 30 16:17:01 <Tiersten> Yep.  Got it
Jun 30 16:17:03 <Tiersten> Finally making sense
Jun 30 16:17:13 <ByronT> now I'm confused
Jun 30 16:17:13 <beewoolie-afk> So, R64 is A23, R81 is A22, and R83 is A21.
Jun 30 16:17:25 <Tiersten> Yep.  Appears so
Jun 30 16:17:48 <dyoung-zzzz> A23 = R64 A22 = R81 A21 = R83
Jun 30 16:17:54 <rwhitby> 010 to 011
Jun 30 16:17:55 <beewoolie-afk> The idea was moving the resistor from R64 to R81 should have made the 0b010 into 0b100
Jun 30 16:17:59 <dyoung-zzzz> so wrongly transposted R83 and R82 yesterday/
Jun 30 16:18:32 <beewoolie-afk> What dyoung just said.
Jun 30 16:18:43 <beewoolie-afk> Uh, oh.  gotta go pack for the weekend.
Jun 30 16:18:44 <beewoolie-afk> ttfn.
Jun 30 16:18:51 <dyoung-zzzz> bye beewoolie
Jun 30 16:21:43 <rwhitby> A23 = R64 A22 = R81 A21 = R83       010 to 011
Jun 30 16:21:52 <rwhitby> is that the consensus?
Jun 30 16:21:57 <Tiersten_> argh
Jun 30 16:21:59 <Tiersten_> yes
Jun 30 16:22:29 <Tiersten_> Removing R64 and R83 are both undocumented but R83 is less undocumented if that makes any sense :)
Jun 30 16:22:36 <dyoung-zzzz> Yes.
Jun 30 16:22:57 <dyoung-zzzz> 011 is documented?
Jun 30 16:23:09 <Tiersten_> Sort of
Jun 30 16:23:11 <rwhitby> No, the Linksys option is undocumented.  Removing R64 goes from one undocumented setting to another undocumented setting.  Removing R83 goes to a documented setting.
Jun 30 16:23:23 <ByronT> what rwhitby said
Jun 30 16:23:25 <dyoung-zzzz> no, it really is documented.
Jun 30 16:23:36 <dyoung-zzzz> page 326 of 25248005.pdf
Jun 30 16:23:37 <Tiersten_> It's documented in that it's documented for the 533MHz part
Jun 30 16:23:44 <rwhitby> 011 is ridgy didge
Jun 30 16:23:46 <dyoung-zzzz> No its a valid option for the 266mhz part.
Jun 30 16:23:54 <Tiersten_> No.  I mean the Linksys one
Jun 30 16:24:00 <Tiersten_> What we're doing is fine
Jun 30 16:24:12 <dyoung-zzzz> Right, 010 is not documented.
Jun 30 16:24:16 <rwhitby> linksys one is not documented anywhere
Jun 30 16:24:16 <Tiersten_> ridgy didge?
Jun 30 16:24:29 <rwhitby> yeah, ridgy didge
Jun 30 16:24:39 <rwhitby> bonza
Jun 30 16:24:47 <rwhitby> happy as larry
Jun 30 16:24:49 <Tiersten_> Anybody got a aussie to brit translator? :)
Jun 30 16:24:57 <rwhitby> she'll be apples
Jun 30 16:25:02 <dyoung-zzzz> A-ok
Jun 30 16:25:06 <dyoung-zzzz> its all good.
Jun 30 16:25:15 <dyoung-zzzz> Bob's yer uncle.
Jun 30 16:25:42 <Tiersten_> So actually our NSLU2s are more in spec :)
Jun 30 16:25:47 <rwhitby> yeah, remove R83 and bob's your uncle
Jun 30 16:26:02 <Tiersten_> Robert is your fathers brother?
Jun 30 16:26:02 <rwhitby> yep, we have simply fixed an error on the Linksys board
Jun 30 16:26:14 <rwhitby> yep, bob's your uncle
Jun 30 16:26:15 <Tiersten_> Neat
Jun 30 16:26:27 <ByronT> weirdo's all
Jun 30 16:26:32 <ByronT> ;)
Jun 30 16:26:36 <Tiersten_> Time to document this anyway
Jun 30 16:26:40 <rwhitby> bunch of galahs
Jun 30 16:26:43 <Tiersten_> Before we forget and have to go through this all again
Jun 30 16:26:55 <jacques> heh
Jun 30 16:28:30 <rwhitby-away> see ya round like a rissole
Jun 30 16:29:14 <ByronT> did he just call us something bad?
Jun 30 16:29:32 <ByronT> if I weren't 12,000 miles away, I might get riled up
Jun 30 16:29:51 <rwhitby-away> no, I would say "seppos" if I wanted to rile you up
Jun 30 16:30:13 <ByronT> heh (something tells me I don't want to know...)
Jun 30 16:30:56 <rwhitby-away> Seppo <- Septic Tank <- Yank <- American :-)
Jun 30 16:31:25 <rwhitby-away> abbreviated rhyming slang
Jun 30 16:31:57 <dyoung-zzzz> Interesting.
Jun 30 16:32:12 <jacques> oh yeah, they did send a lot of cockney criminals to AU didn't they
Jun 30 16:32:30 <rwhitby-away> "Call you on the bone" : Bone <- Dog and Bone <- Telephone.
Jun 30 16:33:00 <jacques> from ocean's eleven I know "barney"
Jun 30 16:33:01 <dyoung-zzzz> thats pretty cool actually.
Jun 30 16:33:12 <dyoung-zzzz> playing with language is fun
Jun 30 16:33:12 <rwhitby-away> "You want some horse on that"; Horse <- Dead Horse <- Tomato Sauce (ketchup)
Jun 30 16:33:14 <jacques> we're in some Barney now
Jun 30 16:33:52 <ByronT> we're keeping rwhitby-away from going to work...
Jun 30 16:34:16 <rwhitby-away> yeah, bye
Jun 30 16:34:50 <NAiL> rwhitby-away: Everything monotone is up?
Jun 30 16:42:03 <Tiersten_> Okay.  Added
Jun 30 16:42:26 <Tiersten> http://www.nslu2-linux.org/wiki/HowTo/OverClockTheSlug
Jun 30 16:42:32 <Tiersten> Feel free to hack to pieces
Jun 30 16:42:48 <ByronT> if I've said it once, I've meant to say it a thousand times.... Tiersten is THE Man
Jun 30 16:43:07 <Tiersten> Yes... That is my surname.  Don't wear it out!
Jun 30 16:48:35 <Tiersten> R65 sets the PCI speed it seems.  Not that we can run the NEC at 66MHz
Jun 30 16:50:21 <rwhitby-away> NAiL: yep mono is up - but you need to start from scratch
Jun 30 16:51:15 <NAiL> I know, syncing now, gonna do patch by patch as soon as I get some caffeine
Jun 30 16:54:10 <NAiL> Man, the backlogs are insane
Jun 30 17:36:14 <Tiersten> I've added a page on the IXP config registers now.  Only cover reg 0 since reg 1 is software controlled
Jun 30 17:37:02 <Tiersten> Squinting at my slug and the bare PCB pics on the wiki haven't been totally clear however and I don't know where 4 of them actually go.  I know what they're set to, just don't know where they are
Jun 30 18:09:13 <Tiersten> I think beewoolie is fairly safe about 110
Jun 30 18:09:38 <Tiersten> The documentation for the IXP465 has a slightly bigger table and it appears 1xx is always the highest speed
Jun 30 18:10:00 <NAiL> Good. My slug is modded the same way.
Jun 30 18:10:02 <Tiersten> There are gaps in our config register that are filled in by the tables in there as well for hardware we don't have
Jun 30 18:10:17 <Tiersten> It's still not recommended anyway :)
Jun 30 18:10:34 <NAiL> No, but mine was modded before we figgered that out :P
Jun 30 18:10:36 <Tiersten> Just on off chance that there will be a NSLU2 with a C0 stepping IXP
Jun 30 18:10:40 <Tiersten> 420
Jun 30 18:10:50 <Tiersten> on the off
Jun 30 18:10:51 <NAiL> I'm getting another slug today, so I'm modding that the "good" way.
Jun 30 18:10:54 <Tiersten> Not that there is a C0 anyway
Jun 30 18:11:02 <Tiersten> How many do you have anyway??
Jun 30 18:11:32 <NAiL> I had two. One died of lightning. Now I'll be back at two.
Jun 30 18:11:55 <Tiersten> ah
Jun 30 18:13:25 <Tiersten> We've got two more combinations to test if somebody is willing :)
Jun 30 18:13:33 <[g2]> Tiersten, are you building 465's ? :)
Jun 30 18:13:45 <rwhitby-web> died of lightning?
Jun 30 18:13:46 <Tiersten> Nope.  I just have the datasheets
Jun 30 18:13:53 <NAiL> yup
Jun 30 18:14:12 <NAiL> It was scorched(sp?)
Jun 30 18:14:19 <Tiersten> How?
Jun 30 18:14:25 <Tiersten> Surge over the powerlines?
Jun 30 18:14:33 <NAiL> yeah
Jun 30 18:14:57 <NAiL> All my other stuff was behind my trusty UPS. So only the UPS blew.
Jun 30 18:15:08 <Tiersten> Still ouch
Jun 30 18:15:21 <Tiersten> We're up to 15 turboslugs which is pretty impressive considering it's only been a day
Jun 30 18:15:37 <Tiersten> True, a lot of them are duplicates but still :)
Jun 30 18:15:40 <NAiL> Consider myself lucky there. It could have kille three PC's, two big monitors, two slugs and more
Jun 30 18:15:59 <Tiersten> yeah
Jun 30 18:16:04 <NAiL> Instead, it killed what was my devslug.
Jun 30 18:16:32 <NAiL> Its memory remains with me ;-)
Jun 30 18:16:34 <Tiersten> Normally a SMPS kills itself when it gets a surge and the device is okay
Jun 30 18:16:43 <NAiL> SMPS?
Jun 30 18:16:43 <Tiersten> How badly scorched is it?
Jun 30 18:16:49 <Tiersten> switched mode power supply
Jun 30 18:16:52 <NAiL> aah
Jun 30 18:17:35 <NAiL> the board was burnt rather badly
Jun 30 18:17:38 <Tiersten> The power supply that comes with the NSLU2 is weird.  You can remove the actual plug bit
Jun 30 18:17:53 <Tiersten> Not that I recommend anybody does that because you end up with a death plug
Jun 30 18:18:07 <[g2]> rwhitby-web, I'm with you on the mid-life speed bump
Jun 30 18:18:26 <NAiL> Tiersten: remove plug bit?
Jun 30 18:18:38 * NAiL checks
Jun 30 18:18:42 <Tiersten> You can remove the actual prongs
Jun 30 18:19:08 <Tiersten> So they can build a single PSU and just put a different set of prongs on it for whatever country they're elling it in
Jun 30 18:19:21 <NAiL> yeah, same as my iPod power sply
Jun 30 18:19:36 <Tiersten> Yeah but you don't get a death plug with the ipod PSU
Jun 30 18:19:49 <Tiersten> That one is designed to take off.  The NSLU2 one isn't
Jun 30 18:19:51 <NAiL> hehe, no
Jun 30 18:20:11 <Tiersten> Unless of course you really do want a plug that has two big exposed contacts :)
Jun 30 18:20:46 <NAiL> naah
Jun 30 18:21:06 <NAiL> Anyways, I'll be back to two slugs this afternoon (*should* be this afternoon)
Jun 30 18:21:57 <NAiL> I'll probably get a couple more later :D
Jun 30 18:22:19 <Tiersten> :)
Jun 30 20:14:39 <NAiL> [g2]: ping?
Jun 30 20:14:44 <[g2]> Pong
Jun 30 20:14:50 <NAiL> low latency
Jun 30 20:15:06 <[g2]> I just happened to be looking in the window
Jun 30 20:15:11 <NAiL> :)
Jun 30 20:15:18 <NAiL> I was wondering about APEX
Jun 30 20:15:24 <[g2]> actually, there's 110V hooked to my nads
Jun 30 20:15:29 <NAiL> haha
Jun 30 20:15:56 <[g2]> it *jump* everytime someone sends [g2]
Jun 30 20:16:05 <NAiL> [g2] :)
Jun 30 20:16:11 <[g2]> so what's up with APEX ?
Jun 30 20:16:22 <NAiL> I was wondering about the advantages
Jun 30 20:16:36 <NAiL> Faster bootup, more free flash right?
Jun 30 20:16:46 <[g2]> that's the plan
Jun 30 20:16:52 <[g2]> I think you can get a faster boot
Jun 30 20:17:10 <NAiL> It's not faster now?
Jun 30 20:17:22 <NAiL> dyoung-overclk: What are you clocking now? :)
Jun 30 20:17:34 <[g2]> well I'm booting by loading the the kernel from within the jffs2
Jun 30 20:17:41 <[g2]> that's slower
Jun 30 20:17:55 <NAiL> ah, yes, it was that issue
Jun 30 20:18:06 <NAiL> I so need to get jtag-enabled
Jun 30 20:18:15 <[g2]> because the bootloader is loading up th jffs2 fs nodes and finding and reading out the /boot/bzImage file
Jun 30 20:18:38 <[g2]> to be safe yes
Jun 30 20:18:51 <[g2]> you've got serial right ?
Jun 30 20:18:57 <NAiL> I'm considering flashing maimedslug anyways ;-)
Jun 30 20:19:10 <NAiL> not on maimedslug, actually... That's a slight problem
Jun 30 20:19:25 <NAiL> If I connect the serial, the slug hangs
Jun 30 20:19:26 <[g2]> hence the maimed part
Jun 30 20:20:29 <[g2]> truth be told I haven't been JTAG enabled on any of my slugs (but one has wires attached but I haven't gotten it to work) I haven't tried real hard either
Jun 30 20:20:40 <NAiL> ah
Jun 30 20:20:58 <[g2]> However, I do have JTAG on the Avila board
Jun 30 20:21:10 <NAiL> yeah
Jun 30 20:21:16 <dyoung-reclk> I'm thinking about jtagging lennerts slug.
Jun 30 20:21:16 <[g2]> and can reflash the bootloader and have on that
Jun 30 20:21:22 <dyoung-reclk> I'm not convinced yet.
Jun 30 20:21:27 <[g2]> dyoung-reclk, that'd be awesome
Jun 30 20:21:33 <NAiL> How hard is getting jtag to work?
Jun 30 20:21:35 <dyoung-reclk> Lennerts Dead Slug.
Jun 30 20:21:48 <[g2]> dyoung-reclk, no need for that 
Jun 30 20:21:56 <dyoung-reclk> Lennerts Working Slug already has JTAG.
Jun 30 20:22:02 <[g2]> that's really cool!
Jun 30 20:22:09 <dyoung-reclk> he cc'd you the message, I'm sure of it.
Jun 30 20:22:14 <[g2]> nod.
Jun 30 20:22:31 <[g2]> NAiL, you built mysql right ?
Jun 30 20:22:41 <NAiL> eah
Jun 30 20:22:44 <NAiL> yeah
Jun 30 20:22:51 <[g2]> does that build cross ?
Jun 30 20:22:57 <[g2]> or it's it a native job
Jun 30 20:23:02 <NAiL> both
Jun 30 20:23:12 <[g2]> does it work cross ?
Jun 30 20:23:20 <NAiL> the cross build had strange paths and stuff, but that's fixed in mtn now
Jun 30 20:23:32 <[g2]> cool
Jun 30 20:23:47 <[g2]> how long did it take to build native ?
Jun 30 20:23:51 <NAiL> I haven't tested the latest build, but the only changes are paths, so I don't think it breaks that hard.
Jun 30 20:24:08 <NAiL> four-five hours I think
Jun 30 20:24:17 <NAiL> I never remember to time stuff like that
Jun 30 20:24:33 <[g2]> so maybe two-three with TurboFatty
Jun 30 20:24:55 <NAiL> I'd guess around three
Jun 30 20:25:15 <[g2]> I'm playing with apache right now on the TF
Jun 30 20:25:22 <NAiL> It'll swap wildly when it compiles stuff like sql_yacc.c, which takes a fair bit of memory
Jun 30 20:25:47 <[g2]> I've got Apache with SSL running
Jun 30 20:26:11 <[g2]> and does openssl cross build fine also ?
Jun 30 20:26:33 <NAiL> Builds fine, no idea whether it works or not. Haven't touched openssl
Jun 30 20:27:00 <[g2]> ok thx
Jun 30 20:27:10 <[g2]> I'm trying to get the LAMP stuff all sorted out
Jun 30 20:27:17 <NAiL> cool
Jun 30 20:27:44 <NAiL>  [g2] [g2] [g2] [g2] [g2] [g2] [g2] [g2] [g2]  <-- How's ya nads? 
Jun 30 20:27:50 <[g2]> hot
Jun 30 20:28:04 <[g2]> was it good for you too ?
Jun 30 20:28:05 <[g2]> :)
Jun 30 20:28:51 <NAiL> Oh yes ;)
Jun 30 20:29:12 <rwhitby-web> still no non-IRC turbosluggers
Jun 30 20:29:32 <NAiL> I can stop hanging on IRC, if that'd help? :P
Jun 30 20:29:33 <[g2]> that'll take a while
Jun 30 20:29:49 * [g2] kicks NAiL 
Jun 30 20:29:55 <NAiL> Ouch
Jun 30 20:30:00 <[g2]> heh
Jun 30 20:30:39 <NAiL> Is anyone "actively maintaining" openslug-uclibc?
Jun 30 20:31:02 <[g2]> I think jbowler-away runs it sometimes
Jun 30 20:31:03 <rwhitby-lunch> nope.  You want a job?
Jun 30 20:31:15 * [g2] needs a job
Jun 30 20:31:20 <NAiL> I'm considering having one uclibc and one glibc
Jun 30 20:31:28 <NAiL> That means I'd need another slug
Jun 30 20:31:32 <NAiL> ;-P
Jun 30 20:32:06 <jbowler-away> I maintain it as much as I maintain the glibc stuff.
Jun 30 20:32:44 <NAiL> jbowler-away: I'm gonna try the same site-changes for uclibc as for glibc, btw. 
Jun 30 20:33:03 <jbowler-away> Yes, that should work
Jun 30 20:33:22 <jbowler-away> I'm not sure yet whether to set up my central file server as uclibc or glibc.
Jun 30 20:33:37 <jbowler-away> If I use glibc I can use NIS, otherwise I have to find something different - openldap.
Jun 30 20:33:40 <NAiL> At the moment, you should choose glibc :-P
Jun 30 20:33:58 <NAiL> Otherwise, all you'd get would be nameless folders 
Jun 30 20:34:51 <jbowler-away> Not an issue yet, because I don't have openldap working (I am, at this moment, running and testing stuff on uclibc)
Jun 30 20:36:23 <NAiL> Maybe I should look into openldap when I've set up my newslug
Jun 30 20:37:16 <jbowler-away> I've only done the most basic of tests so far, but samba has a whole load of schemas for it.
Jun 30 20:37:46 <jbowler-away> I hit a RTFM stop point trying to work out how to persuade tinylogin to use the LDAP database not /etc/passwd
Jun 30 20:38:52 <NAiL> I've realized I need quite a few slugs ;)
Jun 30 20:39:16 <dyoung-web> I have 3.  Barely enough.
Jun 30 20:39:32 <NAiL> I guess I'll wait until I can get some fat ones
Jun 30 20:40:39 <NAiL> Atleast I'll have a two dedicated buildboxes later this weekend :D
Jun 30 20:41:42 <NAiL> If I'm really lucky, I'll get both a slug and my AMD64 box today
**** ENDING LOGGING AT Thu Jun 30 23:59:57 2005