**** BEGIN LOGGING AT Mon Mar 04 02:59:59 2013 Mar 04 04:42:54 I just picked up another n800. What's the best firmware to put on it these days? Mar 04 04:43:32 trash bin Mar 04 04:43:45 that's where I got it from. Mar 04 04:43:58 also where it belongs Mar 04 04:43:59 :/ Mar 04 04:56:20 hi.. good afternoon Mar 04 04:56:36 I'm having trouble browsing some packages at maemo.org Mar 04 04:58:55 for example, when browsing extras-testing-devel Mar 04 04:59:05 using a web browser Mar 04 04:59:23 I can get up to page 39 (letter L of packages) Mar 04 04:59:38 however, page 40 gives an empty response. Mar 04 04:59:46 i.e., this URL: http://maemo.org/packages/repository/list/fremantle_extras-devel_free_armel/?org_openpsa_qbpager_packages_in_repo_page=40 Mar 04 04:59:56 any suggestions? hints? Mar 04 04:59:59 thanks :) Mar 04 05:00:39 sombragris: why not browse via repository.maemo.org? Mar 04 05:00:42 http://repository.maemo.org/extras-devel/pool/fremantle/ Mar 04 05:00:59 or is it because you want to read description of each package? Mar 04 05:02:21 thedead1440: you got it. I'm browsing trying to get a gist of some packages. I'm hunting especially for twitter clients and presence notifications Mar 04 05:02:40 anyway, thanks for the suggestion. Will keep that in mind Mar 04 05:02:51 however, the issue still stands. I cannot get to page 40 Mar 04 05:03:07 i've reported the issue to the sysadmins; since its basically early morning in EU it may take a few hours but they'll look into it Mar 04 05:03:17 great Mar 04 05:03:21 thank you Mar 04 05:03:24 np Mar 04 05:03:28 you know Mar 04 05:03:35 I bought a N900 last week Mar 04 05:03:37 brand new Mar 04 05:03:49 lost mine last April and always wanted another one Mar 04 05:03:50 that's nice :) Mar 04 05:03:54 oh Mar 04 05:04:08 so... i'm looking to rebuild my environment Mar 04 05:04:37 for twitter i think the clients that work are the tweetian backport and twimgo Mar 04 05:04:47 the migration issues put some bumps on that road but I really appreciate the enormous effort people are doing with that Mar 04 05:05:05 will look into tweetian. Mar 04 05:05:08 http://talk.maemo.org/showthread.php?t=88248 Mar 04 05:05:15 try the backport on post #5 of the thread Mar 04 05:05:27 its alpha but looks quite promising Mar 04 05:06:09 as for twimgo, the libqtm11 requirement is blacklisted. Can't get around it. Mar 04 05:06:17 my N900 skills are all rusty now Mar 04 05:06:19 :P Mar 04 05:06:36 tweego by merlin1991 seems ok too: http://talk.maemo.org/showpost.php?p=1285988&postcount=459 Mar 04 05:06:42 I used to use TweeGo back in the day but now it cannot login Mar 04 05:06:51 oh due to API issues Mar 04 05:06:57 running from the console it echoes a 404 Mar 04 05:07:00 :S Mar 04 05:07:22 merlin1991 says on the post that he would update the api to v1.1 as it expired this month but i wonder if he has come around to actually doing it Mar 04 05:07:29 problem is I need to manage two or three accounts, with two or three searches Mar 04 05:07:45 let hope he does. I didn't see any update, neither in testing or devel Mar 04 05:08:06 so far the best stuff I've seen is tweed suite, but it's single use Mar 04 05:08:10 single-user, sorry Mar 04 05:08:51 yeah auto-builder is broken since somewhile so updated packages are not pushed thru... Maybe later when merlin1991 is around you can ask him if he plans to update TweeGo :) Mar 04 05:09:21 anyway. N900 has that thing. It spoils you. Even with these bumps you simply cannot settle down for something else Mar 04 05:09:27 :P Mar 04 05:09:29 ok Mar 04 05:09:32 :D Mar 04 05:09:49 thedead1440: thank you for the advice. Very useful. Will take that into account. Mar 04 05:09:56 np ;) Mar 04 05:10:27 bye now. it's 02.10 (am) here and I need to hit the sheets. ttyl Mar 04 05:10:37 good night Mar 04 12:16:13 Good morning guys! :) Mar 04 12:16:56 (about discussion from yesterday) can someone explain me, why, precisely, COBS > autobuilder idea was abandoned? Mar 04 12:17:40 DocScrutinizer05 said "Nokia pulled plug on it", but I've though that HiFo isn't now forced to follow what nokia said? Mar 04 12:18:22 I just wonder if there is any sense at all in working on bringing autobuilder back again, especially, that almost no one gest his internals, and X-Fade is MIA Mar 04 12:18:46 i do Mar 04 12:18:54 i've built my own autobuilder Mar 04 12:19:03 nice! so please, fix it for maemo ;) Mar 04 12:19:16 they will welcome you with open arms Mar 04 12:19:17 with a bit of x-fade help on what is used in maemo autobuilder Mar 04 12:19:25 thats the point. Mar 04 12:19:41 maemo autobuilder is snafu Mar 04 12:19:50 but that was mostly because i wanted to have similiar setup Mar 04 12:20:16 what is the problem with it exactly? Mar 04 12:20:27 I just wonder, a tiny bit ;) if it could be sane, to redirect people that could be working in autobuilder, into setting up cobs instead, precisely, finishing work on it Mar 04 12:20:56 at this point, it doesn't matter if it will take 2 months or half of a year - ETA for autobuilder getting online is equally unknown Mar 04 12:21:08 my autobuilder is exactly same as maemo, only difference was repository management Mar 04 12:21:09 jacekowski, ask DocScrutinizer05 Mar 04 12:21:12 I see Mar 04 12:21:23 well, DocScrutinizer05 ^ Mar 04 12:22:15 jacekowski, from my very limited knowledge, at some point, autobuilder started to generate wrong hashsums (or stopped regenerating itj for new packages. Oldest confirmed package with such problem is from 10.2012 Mar 04 12:22:28 then, web interface stopped working, but scp was available Mar 04 12:22:37 later, everything stopped qworking all-together ;) Mar 04 12:22:55 wrong hashsums Mar 04 12:22:56 for techynical details, ask people who have access to it, DocScrutinizer05 should know smth about i Mar 04 12:22:59 it* Mar 04 12:23:04 i thought i was just a problem that key has expired Mar 04 12:23:10 jacekowski, nope Mar 04 12:23:11 it* Mar 04 12:23:19 Which A-GPS serve is more preferable ? Mar 04 12:23:27 I don't think supl.nokia.com is running Mar 04 12:23:29 XATRIX, supl.google.com, sadly Mar 04 12:23:31 which would make all the hashsums incorrect Mar 04 12:23:39 jacekowski, hm? Mar 04 12:23:56 interesting, but not all hashes are wrong Mar 04 12:24:00 but Mar 04 12:24:07 maybe only onesa since key expired? Mar 04 12:24:12 yep Mar 04 12:24:51 no idea, I'm just shotting into the wild, please stay tuned for merlin1991 or DocScrutinizer05 apperance, if you could help with autobuilder, it would be useful to the point of blessing Mar 04 12:25:21 although, the problem here is, that even if you generate new key Mar 04 12:25:26 phones will not trust it Mar 04 12:27:52 Once again, can you tell me the good list of repos i have to setup on my device ? I mean ,that repos i can safely hold always enabled ? Mar 04 12:27:53 jacekowski, can be fixed by abusing cssu, probably Mar 04 12:28:01 From where i can update Mar 04 12:28:05 XATRIX, none at the moment ;) Mar 04 12:28:08 otherwise Mar 04 12:28:10 ~mirrors Mar 04 12:28:11 i heard mirror is http://maemo-archive.wedrop.it/ http://talk.maemo.org/showthread.php?p=1315143#post1315143 http://maemo.merlin1991.at/apt-mirror/, or extras-devel.merlin1991.at - for fighting hashsum error Mar 04 12:28:12 O_O Mar 04 12:28:21 (to keep enabled always) Mar 04 12:28:23 Which of i can trust ? Mar 04 12:28:26 (=none) Mar 04 12:28:36 Khm....Khm.. Mar 04 12:28:36 well, if you trust merlin1991, use his repos Mar 04 12:28:39 jacekowski: Estel_ summarized it pretty correctly Mar 04 12:28:56 * Estel_ is surprised - no irony, just got little to no knowledge here Mar 04 12:29:10 jacekowski: right now builder VM is up but totally dysfunct Mar 04 12:29:46 and we actually have NFC what's up with it Mar 04 12:29:52 What's the difference between community-testing/devel and Extras-testing/devel ? Mar 04 12:30:24 jacekowski: we have a maintainer for repo which is merlin1991, we got *nobody* in community with a faintest clue about autobuilder Mar 04 12:31:36 jacekowski: if you would consider to help, pretty please step up, we'd appreciate any help you can give. I can grant access to autobuilder VM for you Mar 04 12:32:09 i don't have time to do it today Mar 04 12:32:18 but i'm happy to help tomorrow Mar 04 12:32:24 we could just make a new autobuilder Mar 04 12:32:25 haha. It's not about today, it's about at all ;) Mar 04 12:32:31 any time you want :-) Mar 04 12:33:10 but, it would be useful to create access to VM nevertheless, yep? after all, tomorrow is just after today Mar 04 12:33:17 * Estel_ blinks Mar 04 12:36:24 i just have to finish a program that has to be ready last friday Mar 04 12:36:45 but i was given the spec for it today Mar 04 12:37:02 sounds familliar ;) Mar 04 12:40:38 Is it ok to install "h-e-n (USB hostmode GUI) ? Mar 04 12:45:03 Also QCPUFreq - to enable Smartreflex and clock to 250-800 mhz Mar 04 12:45:27 I would skip qcpufreq Mar 04 12:45:35 for rest, search TMO Mar 04 12:45:48 btw both assume that you have kernel-power Mar 04 12:46:06 better use kp52 from first post of kp51 thread (yea, sounds logical, I know...) Mar 04 12:46:44 due to kp52 being release rewady, but no way to release it, with autobuilder down Mar 04 12:46:55 XATRIX: h-e-n is ok Mar 04 12:47:12 Estel_: why don't you lile qcpufreq ? Mar 04 12:47:16 DocScrutinizer05: ok Mar 04 12:47:27 ~oc Mar 04 12:47:27 i heard oc is an Optical Carrier, An OC1 has 672 channels or 44.736 Mbit/s Mar 04 12:47:34 ~omap-oc Mar 04 12:47:35 well, omap-oc is http://mg.pov.lt/maemo-irclog/%23maemo.2010-08-01.log.html#t2010-08-01T22:16:05 read that!, or and this http://mg.pov.lt/maemo-irclog/%23maemo.2011-03-11.log.html#t2011-03-11T03:04:11 Mar 04 12:47:43 ~verclocking Mar 04 12:47:48 ~overclocking Mar 04 12:47:49 "OK, listen up. This is your CPU." apt drops the CPU into a hot frying pan. "This is your CPU on overclocking. Any questions?" Mar 04 12:48:40 or in short: despite widely used by those who think it's "1337", OC is deprecated Mar 04 12:50:15 and smartreflex I already elaborated yesterday(?), it's probably not doing much since our kernel is doing same in software what SR does in hw, and it's deprecated by Nokia/TI(?) for introducing instabilities to at least some devices Mar 04 12:51:50 sr did improve battery life somewhat when I tried it. back when it was enabled Mar 04 12:52:03 but some may indeed unstable Mar 04 12:53:53 Testacular - Spectacular Test Runner for JavaScript Mar 04 12:54:00 Fantastic name. Mar 04 13:06:35 almost as good as fapman Mar 04 13:16:23 What can you advice me to install to be able to dynamically overclock/downclock my CPU + smartreflex ? Mar 04 13:27:55 Is it a good idea to place a thing, that will show my current battery level in percentage or whatever ? Mar 04 13:28:07 Will it drain more energy ? Mar 04 13:47:57 ? Mar 04 13:49:56 HAM is freaking sloooooow!! :( Mar 04 14:12:27 orly? Mar 04 14:13:13 XATRIX: adbice: do not mess with cpuclock and SR Mar 04 14:14:03 showing current batterylevel is useless for more than what battery statusbar applet already shows you Mar 04 14:14:31 and yes, it most likely will drain additional energy from bat just for showing redundant info Mar 04 14:14:47 some of those desktop applets are notorious for doing so Mar 04 14:15:17 HAM gets the slower, the more large repositories you got enabled Mar 04 14:15:39 with extras-devel it takes ~7min to build initial screen Mar 04 14:16:13 ... and 7 minutes again each timeyou actually do anything Mar 04 14:16:59 :( crap Mar 04 14:18:41 originally they planned to name it "zen-teacher" ;-) Mar 04 14:19:00 Why are you so against SR and oclock ? Mar 04 14:19:22 because it's useless and may cause instabilities and damage Mar 04 14:20:28 DocScrutinizer05, HAM is slow, period. Mar 04 14:20:32 e.h HAM could probably get speed up to 6 instead of 7 min, at cost of your device defect after 500h instead of 10000h Mar 04 14:20:46 e.g. Mar 04 14:21:06 it could be argued that it's best used only for 1) initial setup of Extras-* repos, and 2) system upgrades Mar 04 14:21:40 and according to my former explanations that our kernel does cpu clock and voltage regulation in sw, SR is probably absolutely useless Mar 04 14:21:42 * kerio uses it for neither of those Mar 04 14:22:33 kerio: you probably shouldn't join discussions when it comes to teaching newcomers how to use N900 Mar 04 14:23:26 * kerio goes back in his corner, quietly sobbing Mar 04 14:23:30 you're trying to teach how to do a wheely to a bike learner Mar 04 14:25:33 You think overclock of ARM harms the SoC ? Mar 04 14:25:44 It's not a usual overvoltage clocking Mar 04 14:25:48 It's a softclock Mar 04 14:26:11 Like if i would rise the multiplier of Intel/AMD CPUs Mar 04 14:26:26 overclocking of OMAP *massively* reduces expected lifespan Mar 04 14:26:33 I'm not clocking the system bus Mar 04 14:26:40 Hm Mar 04 14:26:42 DocScrutinizer05: don't we have a datasheet regarding that? Mar 04 14:26:51 ~omap-oc Mar 04 14:26:52 somebody said omap-oc was http://mg.pov.lt/maemo-irclog/%23maemo.2010-08-01.log.html#t2010-08-01T22:16:05 read that!, or and this http://mg.pov.lt/maemo-irclog/%23maemo.2011-03-11.log.html#t2011-03-11T03:04:11 Mar 04 14:27:38 second link Mar 04 14:27:45 XATRIX: electromigration, etc. Mar 04 14:29:14 Hurrian: maybe i say bullshit, but it's not a cell , to accumulate electrons... Mar 04 14:29:17 IMHO Mar 04 14:31:19 XATRIX: well, whatever you want to do with your device, we're not stopping you. It /is/ the N900 after all. Mar 04 14:31:40 Just don't come running to the forums when 3 = 1 + 1. Mar 04 14:32:06 or, more likely, when SIGSEGV Mar 04 14:32:09 I think all of the ARM7 CPUs are the same... The same core... The difference is only in a percent of so called 'bad' CPUs on a market Mar 04 14:32:15 SIGSEGV ALL THE THINGS Mar 04 14:32:32 all OCing is usually worthless anyway, since bottleneck is RAM (or lack of) and swapping Mar 04 14:32:37 The all can run at the highest clock, but all can give different results, depending on how 'clean' was it built Mar 04 14:32:47 and.. completely awful code (ham) Mar 04 14:33:30 XATRIX: that's definitely not true Mar 04 14:33:38 Hm , sure ? Mar 04 14:33:46 they get selected at fab Mar 04 14:33:53 so called binning Mar 04 14:33:59 XATRIX: ARMv7 Cortex-A8 is most definitely not the same with Cortex-A15 ;) Mar 04 14:34:07 Yes, the most broken costs cheap, the better costs higher Mar 04 14:34:44 Hurrian: nope, i mean just for example, AMD Athlon 3200+ = AMD Athlon 5600+ Mar 04 14:35:06 But 3200 has much more 'broken' transisters Mar 04 14:35:24 XATRIX: To take from your example, it's like unlocking shaders on a recent Radeon card. Mar 04 14:35:41 If you do happen to unlock shaders on a partially-defective Radeon, it's bricked. Mar 04 14:35:46 so, they effectively sell people trashbin-quality omaps <.< Mar 04 14:35:55 I forgot the word, for a term. The 'way' on machinery plant, where the cars are built Mar 04 14:36:18 Hurrian: Sure, that's what i'm talking about ! Like with shaders! Mar 04 14:36:41 It's just the same GPU as the top most. Mar 04 14:37:02 well, uhh, yeah, see, the problem is, 600 MHz is pretty near the top of line line OMAP 34xx series SoCs TI was shipping at the time Mar 04 14:37:24 But it has locked shader 'faclilities' Mar 04 14:37:37 the only one higher than that is the slightly later-manufactured OMAP 35xx series, the highest of which are 720 something MHz Mar 04 14:37:50 *note: OMAP 34xx = 35xx. Mar 04 14:38:02 Maybe, i don't know :( Mar 04 14:38:15 XATRIX: They're definitely not made to go at 1GHz, although they can do so. Mar 04 14:38:53 It's like redlining an engine - you can do so, still won't stop your engine knocking after a few hours of doing so. Mar 04 14:39:28 i remember old celeron 266:s (without L2 cache), they would go 266->400 without a fan, 266->450 if you add a small fan, and i had one running for about a decade Mar 04 14:39:56 yea Mar 04 14:40:19 I have OCed my Phenom II 965 3.4Ghz to 3.8 (each core) Mar 04 14:40:43 So, my approx profit is ~+1.5Ghz Mar 04 14:41:04 well, cant really count that way.. % Mar 04 14:41:04 And it will not go down earlier then i will upgrade my PC :) Mar 04 14:41:20 s/then/than Mar 04 14:41:30 XATRIX: performance gain on overclocked processors isn't linear, though Mar 04 14:41:42 Hurrian: as i said, apporximately :) Mar 04 14:41:54 memory is still awfully slow on modern pcs Mar 04 14:42:26 Yep Mar 04 14:42:29 jaska: 3000 MHz DDR3 isn't cheap. Mar 04 14:43:00 + 30-45% price for ECC Mar 04 14:43:16 Btw, is it ok, to install kernel-module-ntfs ? Mar 04 14:43:37 XATRIX: use ntfs-3g instead. Also, who uses ECC on a desktop system? Mar 04 14:43:52 i used to use ecc on desktops until they made it require a xeon :( Mar 04 14:44:11 Noone, but who mentioned a desktop system talking about DDR3 , last minute ? :) Mar 04 14:44:20 I'm running a Nehalem Xeon, but I can't be assed to put down that much dosh on ECC RAM. Mar 04 14:44:59 dunno how much 64 gigs of ddr3 ecc is. Mar 04 14:45:28 jaska: Well, a regular 1x8GB DDR3 DIMM is $50. Mar 04 14:45:49 So, 1x8GB of DDR3 ECC should be around $80 or so. Mar 04 14:45:57 i paid around 500 eur for 8x8 of ddr3-2133 Mar 04 14:46:32 jaska: Corsair Dominator Platinums? Mar 04 14:46:41 nah, some kingston set Mar 04 14:46:48 Kingston Value RAM ) Mar 04 14:46:48 Damn, that's expensive. Mar 04 14:46:54 hyperx something Mar 04 14:47:12 I don't think Kingston is a bad RAM Mar 04 14:47:22 It's quite good enough Mar 04 14:47:22 jaska: HyperX blu? Mar 04 14:47:38 beast or somesuch Mar 04 14:48:09 speaking of ValueRAM, I need to buy 4x4GB of DIMMs for my system. Mar 04 14:48:46 Really don't feel like paying twice that for 3x8GB of DIMMs. Mar 04 14:49:31 i wish i could get more ram on the n900 :( Mar 04 14:49:52 Re-solder it Mar 04 14:50:17 XATRIX: You haven't seen how thin that POP RAM is, have you Mar 04 14:50:43 No Mar 04 14:51:11 It's like three A4 sheets stacked. Mar 04 14:51:43 Oh, and you have to be extra careful not to scratch or let any solder on the core in the middle, under the RAM. Mar 04 14:52:50 the amount of memory bugs me far more than the low clockspeed Mar 04 14:53:18 You don't have to Mar 04 14:53:26 Use the hi-temp fan Mar 04 14:53:56 Warm it from the other side, and put the chip from the top Mar 04 14:54:05 It will fit as it was there from the beginning ;) Mar 04 15:03:11 ddr2 is a nightmare prices :/ Mar 04 15:03:15 hi all Mar 04 15:03:39 Yea :) Mar 04 15:03:56 It's even more expensive than DDRI Mar 04 15:04:05 What memory chips is in N900 ? Mar 04 15:04:54 yea more strange that ppl and shops should sold it not keep them with high prices :/ Mar 04 15:06:03 i have ddr2 motherboard and hunting to "human" ddr2 prices... Mar 04 15:06:41 anyway the point is that SoCs / OMAP seems to behave different than desktop PC CPU Mar 04 15:06:42 XATRIX: nand flash 256? Mar 04 15:07:14 Yea Mar 04 15:07:18 and it's already "Overclocked" by fab settings allowing it to go to 600MHz Mar 04 15:07:21 Is it DDR2 or 3 ? Mar 04 15:08:04 there's been a warning from Nokia engineer that you MUST NOT lock it to 600MHz even, otherwise you will fry the N900 Mar 04 15:08:11 DocScrutinizer05: heh... the initial clock was ~0 Mhz ? ;) Mar 04 15:08:35 no, the stabdard sustained max clock is 500MHz Mar 04 15:08:42 standard Mar 04 15:09:08 600 is already "overclocking" it, which is only used for busts Mar 04 15:09:21 bursts* Mar 04 15:09:41 It's not so fast to call it "bursts" Mar 04 15:09:52 for this cpu it is Mar 04 15:10:02 it reduces lifetime of chip to 15% of original design lifetime Mar 04 15:10:18 sad that nokia cant make chips radiators from phone cover :p Mar 04 15:10:49 lifetime of CPU @ 500MHz = 100%, @600 = 15% !!! Mar 04 15:11:09 DocScrutinizer05: thats freq :/ Mar 04 15:11:14 guess what it is @800, or even at 1000 Mar 04 15:11:33 BB runs at higher frequencies Mar 04 15:11:38 i hear thatlife is smaller but why that dramatically? Mar 04 15:11:49 with exactly same chips Mar 04 15:12:02 with just bit more rigorous selection Mar 04 15:12:16 I respect you're experience, and always note what you say. But touching the 600 = 15% - i don't believe it Mar 04 15:12:22 different binning Mar 04 15:12:41 15% seems bit drastic Mar 04 15:12:53 XATRIX: well, that's what is clearly stated in datasheet Mar 04 15:12:55 bb is running at 800 iirc Mar 04 15:13:04 It's not goes overvolt, it doesn't provide too much Watts for +100Mhz.. It can't be almost dead at the beginning Mar 04 15:13:09 ~omap-oc Mar 04 15:13:09 rumour has it, omap-oc is http://mg.pov.lt/maemo-irclog/%23maemo.2010-08-01.log.html#t2010-08-01T22:16:05 read that!, or and this http://mg.pov.lt/maemo-irclog/%23maemo.2011-03-11.log.html#t2011-03-11T03:04:11 Mar 04 15:13:20 I was rinning 800, and still looks good enough Mar 04 15:13:21 XATRIX: +100Mhz means a lot more current Mar 04 15:13:40 XATRIX: current depends on switching frequency Mar 04 15:13:47 jacekowski: current != voltage Mar 04 15:13:49 http://focus.ti.com/lit/ds/symlink/omap3525-hirel.pdf page 122 3.3 Recommended Operating Conditions Mar 04 15:13:57 XATRIX: more current == more power Mar 04 15:14:05 as in, more watts Mar 04 15:14:09 negative Mar 04 15:14:16 P=UI Mar 04 15:14:26 if I increases P increases Mar 04 15:14:29 because U stays constant Mar 04 15:14:42 I had my Centrino processor, it gave 25W @1.6Ghz. I have my next one 2.13 also gave 25W Mar 04 15:14:51 The same voltages Mar 04 15:15:03 different processors Mar 04 15:15:04 but different devices Mar 04 15:15:21 architecture of chip also make big difference Mar 04 15:15:26 that's why the select/bin chips at fab Mar 04 15:15:32 arch in nm Mar 04 15:15:39 in typical FET device there is a small "capacitor" inside each transistor Mar 04 15:16:05 that capacitor has to be charged/discharged and that's where the heat comes from Mar 04 15:16:13 if you increase frequency you increases losses there Mar 04 15:16:21 and you increase switching loses Mar 04 15:16:25 ok sorry, I've been wrong Mar 04 15:16:33 it's not 15% but 20% Mar 04 15:16:50 so let's say at 500MHz you have 5W of power Mar 04 15:16:55 lol bad DocScrutinizer05 bad :p Mar 04 15:16:58 Maybe 100%-20% = 80% lifetime, i can agree Mar 04 15:17:01 joke ofc Mar 04 15:17:05 no Mar 04 15:17:06 at 600Mhz you will have at least 6W Mar 04 15:17:23 it's 50k DOWN *TO* 10k Mar 04 15:17:30 jacekowski: yes, that makes sens Mar 04 15:17:40 NOT down *by* 10k Mar 04 15:17:44 in practice it will be more like 6.5-7W Mar 04 15:18:03 Yes, the overhead Watts Mar 04 15:18:12 once you reach chip limits power increases a lot more due to other effects Mar 04 15:18:27 As i actually told Mar 04 15:18:41 But it still can't be 10% lifetime went down from 100% Mar 04 15:18:45 yes it can Mar 04 15:18:55 just with 10%overclock Mar 04 15:19:05 it's not 10% Mar 04 15:19:08 it's 20% overclock Mar 04 15:19:25 and it's worst case scenario Mar 04 15:19:27 if your steel rope cam lift 1000kg at 20°C and 800kg at 500°C, how much can it lift at 1000°C? Mar 04 15:19:49 50 ? Mar 04 15:19:53 0 Mar 04 15:20:02 depends Mar 04 15:20:03 DocScrutinizer05: 600 kg Mar 04 15:20:14 before it melts Mar 04 15:22:15 and 10k hours is still around a year Mar 04 15:23:28 yes, but then, that's without any overclocking as they call that nonsense Mar 04 15:23:58 how much will that rope lift at 1200°C? Mar 04 15:24:12 at 1400°, 1500° ? Mar 04 15:24:50 you said 10k hours at 600Mhz Mar 04 15:25:07 we have official numbers that tell us 500MHz->600MHz == 100%->20% lifetime Mar 04 15:25:39 which is still perfectly acceptable considering that phone will sleep more than 70% of the time Mar 04 15:25:54 up to you to guess what it will be for 800Mhz Mar 04 15:25:59 so in reality it means 3 years in worst case scenario Mar 04 15:26:39 or even 1000Mhz which some "leete overclockers" suggest would work "just fine for everybody" Mar 04 15:27:14 bottom line: OC is deprecated and usually not worth it Mar 04 15:28:49 and if let's say we could do 1h at 1GHz Mar 04 15:29:06 it's probably going to last longer if we run for 1s at a time at 1GHz Mar 04 15:29:27 I *could* figure a specially patched mplayer that overclocks to 800 for very short periods of time, occasionally, to avoid framedropping during complex transitions of encoding Mar 04 15:29:28 so 1s sleep 1s 1Ghz is probably going to last longer than 1h Mar 04 15:29:45 a lot of this is statistical, I overclocked two celeron 300's to 450MHz. One of which is pretty much dead after around 3 years. Other one is still perfectly fine but... still long since obsolete. YMMV. I never knew the omap was so fragile... Mar 04 15:30:19 eccerr0r: those devices are pushed quite far and have no heatsinking at all Mar 04 15:30:55 eccerr0r: again, we're not talking about celeron but about OMAP, and the figures as of above are hard numbers from datasheet Mar 04 15:31:49 bear in mind that TI will cover their ass Mar 04 15:32:07 and that number probably means that after 10k hours 1 device have failed Mar 04 15:32:18 if reliability drops *that* quickly... must be something that intel does and ti does not... :\ Mar 04 15:32:21 no, 17 of 1000 Mar 04 15:32:51 that number doesn't mean a lot unless we can get full distribution Mar 04 15:33:02 it's about embedded, eccerr0r Mar 04 15:33:05 because it may be that 90% of devices will last 30k hours at 600MHz Mar 04 15:33:08 the question is... what about samsung and other chips. Mar 04 15:33:11 eccerr0r: same Mar 04 15:33:16 err... manufacturers. Mar 04 15:33:19 embedded is a whole different bag of BS Mar 04 15:33:40 i should overclock my atom chip and see when it will die Mar 04 15:33:50 eccerr0r: for ARM normal "idle" temperature is around 60C when running Mar 04 15:33:52 the whole chip is smaler than celeron die Mar 04 15:34:06 eccerr0r: and when you load it it's twice as much Mar 04 15:34:15 :nod: Mar 04 15:34:28 which causes thermall stresses that don't help Mar 04 15:34:39 exactly Mar 04 15:34:46 and then you have electromigration Mar 04 15:34:58 ~wiki electromigration Mar 04 15:35:01 At http://en.wikipedia.org/wiki/Electromigration (URL), Wikipedia explains: "{{More footnotes|date=July 2010}} 'Electromigration' is the transport of material caused by the gradual movement of the ions in a conductor due to the momentum transfer between conducting electrons and diffusing metal atoms. The effect is important in applications where high direct current densities are used, such as in microelectronics and related structures. As the structure ... Mar 04 15:35:20 celeron running at 450MHz will have reasonably stable temperature Mar 04 15:35:46 and those chips used to be made in much bigger process Mar 04 15:35:56 the l2-cacheless old celeron was a slug without the oc :) Mar 04 15:37:08 XATRIX: so again, do not OC! it's not as leete as you might think Mar 04 15:37:16 my n900 can't get to 850 Mar 04 15:37:22 Alright, i won't Mar 04 15:37:32 i think Mar 04 15:37:43 kerio: mine does 850 pretty well, and stable as hell Mar 04 15:37:46 thing is, rpi for example Mar 04 15:38:24 normal clock for that chip is 700Mhz and people run it at close to 1GHz Mar 04 15:39:09 What's the name of the app, that displays cpu load/memory in tray ? Mar 04 15:39:31 and so far there has been no reports of premature failures Mar 04 15:41:29 i trust and also dont oc my n900... Mar 04 15:41:49