**** BEGIN LOGGING AT Tue Mar 19 02:59:58 2013 Mar 19 03:08:40 Russ: what schematic capture and layout are you using? Mar 19 03:12:36 gschem/geda pcb Mar 19 03:15:16 Russ: i need to retest with that Mar 19 03:15:45 Russ: i've been using kicad, but i really don't like how the applications aren't tightly integrated Mar 19 03:15:48 prpplague, it takes some getting used to, but I've used it for more than 12 years Mar 19 03:16:02 beer o clock Mar 19 03:16:08 mranostay: ding! Mar 19 03:16:22 Russ: i just cant seem to get out of using orcad+pads Mar 19 03:16:35 i am hoping to get moved over to diptrace Mar 19 03:17:02 are we playing what our favorite circuit CAD is? Mar 19 03:18:45 mranostay: trying to evaluate some options for CCO Mar 19 03:18:57 #eagleallthings Mar 19 03:21:10 Crofton: !!!!! Mar 19 03:21:19 I am back for a bit Mar 19 03:22:11 prpplague: btw i'm totally making a Hello Kitty limited edition Bacon Cape :P Mar 19 03:22:29 mranostay: hehe Mar 19 03:22:54 * prpplague wonders what just happened Mar 19 03:23:04 prpplague: ironically i did a IIO patch this weekend Mar 19 03:23:10 to expose the ADC values Mar 19 03:23:10 big boom and the shook Mar 19 03:23:32 prpplague: we need a real driver for the shift register? Mar 19 03:23:37 or SPIdev fine? Mar 19 03:24:17 spidev is ok for now Mar 19 03:24:34 * prpplague wonders what that boom was Mar 19 03:24:53 i'm sure kridner will javascript this #$@$# up :) Mar 19 03:25:25 sssh Mar 19 03:25:33 don't speak the evil ones name Mar 19 03:25:39 kridner? Mar 19 03:26:06 hehe either one of the J's in your statement Mar 19 03:26:32 i hope he reads these logs Mar 19 03:26:46 JAVASCRIPT SUCKS! Mar 19 03:26:50 *whew* Mar 19 03:27:51 * mranostay plays find the eagle part Mar 19 03:34:31 interesting, seems diptrace works perfectly (or so it seems) under wine Mar 19 03:35:26 prpplague: never trust things that just work Mar 19 03:37:05 hmm Mar 19 03:37:13 * prpplague wonders what formats eagle exports to Mar 19 03:39:15 * prpplague decides to test diptrace under wine for 30 days Mar 19 03:44:45 prpplague: I just did a design in diptrace on Win7 - went very nicely. Mar 19 03:45:14 haven't tried it w/ Wine tho. Mar 19 03:50:06 wait windows still exists? Mar 19 03:50:32 emeb_mac: oh ? Mar 19 03:50:47 emeb_mac: any design files you could share me to play with for testing? Mar 19 03:51:32 prpplague: sorry, no - this is for a customer. that's why I used diptrace - it's their internal std. Mar 19 03:52:27 emeb_mac: ahh Mar 19 03:52:47 emeb_mac: question, what kind of reports can you get on the pcb layout stuff? Mar 19 03:53:45 prpplague: what sort of reports? you looking for things drill lists, net lengths, etc? Mar 19 03:54:27 emeb_mac: yea net lengths, i.e. for groups of differential pairs Mar 19 03:55:18 prpplague: I know that dt reports net lengths "live" when you mouse over them. Haven't tried matching tho. Mar 19 03:55:53 I'll fire it up and see... Mar 19 03:57:52 emeb_mac: be interesting how it reports something like a group of fpdlink pairs Mar 19 03:58:12 prpplague: yeah that would be handy Mar 19 03:58:33 * prpplague tests orcad9 under wine Mar 19 03:59:43 emeb_mac: hehe btw, if you do a virtual box install of windows xp, install the 30 day eval of diptrace, then make a image backup, you can use it forever if you wanted to, hehe Mar 19 04:00:31 prpplague: naughty naughty. ;) Mar 19 04:02:50 emeb_mac: i have purchased a full copy, but i found that out while i was doing the eval Mar 19 04:03:27 prpplague: I've been thinking about getting a copy too. Mar 19 04:12:39 doh Mar 19 04:12:51 orcad installs but you can't place any parts Mar 19 04:12:54 doh doh Mar 19 04:14:52 ka6sox seems to has gotten closer Mar 19 04:15:28 * prpplague really needs to backup data off this machine and reload it Mar 19 04:16:42 ka6sox seems to has bit farther but not too far Mar 19 04:33:44 prpplague: gg sparkfun they don't have the small slider part Mar 19 04:40:08 woot, I feel like a newb.. but I JUST got a cmake project that I hacked up to cross compile Mar 19 04:40:31 thurbad: beer time! Mar 19 04:40:34 woot Mar 19 04:40:40 totally Mar 19 04:42:01 it may have taken me a couple hours to figure it out but the project took longer than that to compile natively a single time Mar 19 04:44:31 actually most of that time was spent foolishly trying to get it into eclipse Mar 19 05:09:28 ethically compiled Mar 19 05:12:34 <_av500_> yarn Mar 19 05:13:17 i've just had some allegro/orcad files converted to eagle Mar 19 05:13:21 stupid clock skew.... stupider me for not having the locale set on my beagle Mar 19 05:13:36 <_av500_> mrpackethead: eagle ftw Mar 19 05:14:16 For the Win? Mar 19 05:14:24 <_av500_> ftl then :) Mar 19 05:14:32 ftm actually Mar 19 05:14:56 <_av500_> everybody is happy Mar 19 05:15:15 theres some guys who do a conversion service, and its pretty good Mar 19 05:15:28 * _av500_ teleports to work Mar 19 05:26:16 even more badass Mar 19 05:32:19 i'm not really a fan of eagle, but i'm less of a fan of moving now Mar 19 05:32:20 :-( Mar 19 05:47:18 JAVA4EVER! Mar 19 05:48:46 now you're trolling Mar 19 05:49:22 if he wants cross platform... Mar 19 05:49:23 damn teleporters fluxxed, had to reroute via Jupiter station Mar 19 06:37:54 av500: hell of a reroute Mar 19 07:15:07 [kernel] koenkooi pushed 1 new commit to 3.8: http://git.io/Kv9URw Mar 19 07:15:07 kernel/3.8 4bfe375 Koen Kooi: 3.8: fix email address in USB NOP patch... Mar 19 07:31:06 just learned the -j compiler option... Mar 19 07:31:10 <- happy camper Mar 19 07:34:22 :) Mar 19 07:34:39 -j is addictive Mar 19 07:34:46 it makes you shop for more cores Mar 19 07:35:09 heh, I don't thinks macbooks come with more than 4 yet Mar 19 07:35:28 but I'm considering dedicating a 3rd one to my vm ~.~ Mar 19 07:36:59 had a coworker that tried to dedicate 6 cores to his vm... his host only had 4 Mar 19 07:37:09 :) Mar 19 07:39:48 of course I /could/ leave the macbook at home and get another pc dedicated to linux Mar 19 07:40:15 unthinkable Mar 19 07:40:46 yeah, can't get to productive... they'd come to expect it :P Mar 19 07:47:59 gcc -j or make -j ? Mar 19 07:48:34 ls -j Mar 19 07:49:13 I did some tests with xz -T0 and telling the VM to show 8 cores to linux was faster than 4 cores Mar 19 07:49:22 on a quad i7 with 8 threads Mar 19 07:49:23 make Mar 19 07:49:55 I do wish parallels vm would allow more than 8GB of ram for the vm Mar 19 07:51:38 ht Mar 19 08:10:36 koen: you use vm for building your roms? isn't quite a lot of power burned in the virtualization layers? Mar 19 08:12:31 mainly the io is considerably slower for me, no matter if it's on SSD Mar 19 08:15:42 is this a dog owners forum? Mar 19 08:16:46 rossrossau: no Mar 19 08:17:40 no? I've a dog Mar 19 08:18:02 so you mean this is a phorum? Mar 19 08:18:27 since there's a lot of troll, I think so Mar 19 08:18:34 s/troll/trolls/ Mar 19 08:20:42 and he didn't said yet, if his dog is feed with granules or with 5V DC supply Mar 19 08:21:46 as he can't read the topic, I barely think he knows what "5v dc" means Mar 19 08:42:01 ynezz: I don't build ROMs Mar 19 08:42:10 ynezz: I build filesystems you can write to Mar 19 08:42:16 read-only is a waste of time Mar 19 08:42:41 ynezz: yes, IO is a killer, that's why I also have native build machines Mar 19 08:42:56 ynezz: but my linux build machines are a core2quad and dualcore i7 Mar 19 08:43:03 the vm a quad i7 Mar 19 08:43:27 koen: I doubt he meant ROM in the ROM Sense Mar 19 08:43:30 so for cpubound stuff I tend touse the VM Mar 19 08:43:31 more in the XDA sense :) Mar 19 08:43:47 I see Mar 19 08:44:03 I don't ransom angstrom binaries for donations Mar 19 08:46:01 koen: AngstROM ? Mar 19 08:46:05 :) Mar 19 08:46:46 KoenogenMod... Mar 19 08:53:43 cheese Mar 19 09:07:19 my beaglebone isrunning with angstrom, how can I start telnet server? Mar 19 09:08:52 sshd runs by default Mar 19 09:13:53 success, thanks. Mar 19 09:35:50 telnet.. geeper hav'nt really used that for a long time Mar 19 09:35:58 but in the bad old days.... Mar 19 09:40:42 I used it yesterday :) Mar 19 09:40:48 on this DSP evm :) Mar 19 09:49:42 yeah, i guess there still a place to use it Mar 19 09:50:13 still handy for checking mail servers Mar 19 10:12:34 how much current can a bone GPIO do? Mar 19 13:13:25 bored.. Mar 19 13:13:31 +1 Mar 19 13:18:21 heh Mar 19 13:18:30 I searched for 'micro SD' on bol.com Mar 19 13:18:41 it returns a filter where you can filter on manufacturer Mar 19 13:18:47 'archos' is one of them Mar 19 13:18:52 they sell micro SD cards Mar 19 13:18:56 with a tabled attached :) Mar 19 13:25:39 tablets got so cheap that you get them now as gift for buying SD cards? Mar 19 13:25:44 :) Mar 19 13:25:48 heh Mar 19 13:40:39 is purple squared a purple so purply it would make Prince cry? Mar 19 13:42:10 XorA: purple squared: http://images3.wikia.nocookie.net/__cb20110707190514/deadliestjokes/images/2/2c/Barney.jpg Mar 19 13:42:40 I always just want to punch that! Mar 19 14:23:34 can I power the beagle with a 5v bench power supply on the P9 expansion header VDD_5V? Mar 19 14:24:42 jackmitchell: or sys_5v, I always forget which one is directly connected to the DC barrel jack Mar 19 14:27:37 koen: SYS_5V goes into the TPS65217B and VDD_5V goes into NCP349, so it should be SYS_5V, right? Mar 19 14:28:03 I think so Mar 19 14:28:15 easy enough to check Mar 19 14:28:22 power over usb and measure the pins Mar 19 14:28:34 the pins that show 0V are the ones you want Mar 19 14:28:42 ah, ok, so the USB power doesn't come out over those pins Mar 19 14:30:52 koen: VDD_5V Mar 19 14:44:01 so you were right to start with :) Mar 19 14:48:31 P L A G U E Mar 19 14:49:25 THE P L A G U E Mar 19 14:49:39 I stand corrected Mar 19 14:49:43 and also "the alan" Mar 19 14:49:50 hehe Mar 19 14:50:07 hi prpplague Mar 19 14:50:34 panto: greetings Mar 19 14:50:51 trolls! Mar 19 14:51:08 you say it like it's a bad thing Mar 19 14:51:29 <_troll_> ACTION  Mar 19 14:54:32 does anyone know of a reason that malloc would return a value such as 0x01? Mar 19 14:54:56 clearly that's a bad pointer Mar 19 14:54:58 bugs Mar 19 14:55:07 ticket closed? Mar 19 14:55:17 lol, thanks Mar 19 14:55:34 thurbad: you've probably messed up malloc's data structures Mar 19 14:55:43 by overwriting something at a previous time Mar 19 14:55:45 or dynamic linking is fucked Mar 19 14:56:02 once malloc/free show symptoms, the damage is long done. Mar 19 14:56:16 thurbad, valgrind... Mar 19 14:56:19 +1 Mar 19 14:56:24 yeah, that Mar 19 14:56:36 +1 valgrind Mar 19 14:56:37 valgrind fucked up malloc? Mar 19 14:56:41 *g* Mar 19 15:00:02 could mixing new and malloc cause that? Mar 19 15:02:54 new is just a fancy wrapper around malloc Mar 19 15:03:12 not necessarily Mar 19 15:03:23 stuff allocated with new has to be freed with delete Mar 19 15:03:35 thurbad: did you do "malloc = new"? Mar 19 15:03:53 mru: i didnt say it doesnt do anything additional Mar 19 15:04:06 it doesn't have to call malloc Mar 19 15:04:14 hmm.. right Mar 19 15:04:15 it could use a completely separate allocator Mar 19 15:04:22 most compilers just call malloc Mar 19 15:04:29 but that should never be relied on Mar 19 15:04:44 but then it should work well together with malloc... Mar 19 15:04:59 if not, the compiler is broken... but that would be expected from c++ anyways Mar 19 15:05:00 sometimes mixing works, yes Mar 19 15:05:16 but you still shouldn't do it Mar 19 15:05:28 of course you shouldn't use c++ in the first place Mar 19 15:05:58 delete(c++); Mar 19 15:07:11 there's always the delete vs delete[] issue too. My understanding is that calling the former instead of the latter (when the latter is required) will simply leak a few bytes (ie: not screw up the allocation structures) Mar 19 15:07:26 most likely, an array was overwritten. Mar 19 15:07:35 sounds very implementation-specific Mar 19 15:08:08 yes, I'm talking about gcc, which I assume thurbad to be using. Mar 19 15:08:31 and I could be wrong. I read that a long time ago and don't remember where. Mar 19 15:09:45 actually, a delete when delete[] is needed, should fail... Mar 19 15:09:51 eg Mar 19 15:09:53 but i doubt all c++ compilers do that Mar 19 15:10:41 WTF is delete[]? and surely that breaks the whole principle :-( Mar 19 15:10:58 XorA, welcome to C++, where no-one knows the whole language Mar 19 15:11:07 delete[] is an array of objects Mar 19 15:11:38 you better use the correct form... Mar 19 15:11:40 C++ always was the language for people who refused to program properly in C, seems they are still at it Mar 19 15:12:04 just seen in a channel next door: Bar Foo::bar = [](){ return 42;}(); (no, this is not perl) Mar 19 15:13:03 so because C++ still allows C style arrays which is dumb in itself they need a spacial delete operator to handle it? Mar 19 15:13:27 XorA: nope Mar 19 15:13:38 XorA: delete[] is the antagonist to new[] Mar 19 15:13:53 XorA: i guess strubstrub wanted to make it symetric Mar 19 15:14:06 stroustup is an idiot! Mar 19 15:14:38 obligatory: http://harmful.cat-v.org/software/c++/I_did_it_for_you_all Mar 19 15:14:50 I don't see an issue with the [] notation to denote it's an array of objects? Mar 19 15:14:53 strstr()? Mar 19 15:15:17 av500: that's c! the c++ equivalent would be stringstring() Mar 19 15:15:37 stroustru() Mar 19 15:15:53 C! comes one char after C# :-D Mar 19 15:15:54 * KotH votes for strubstrub() Mar 19 15:16:02 +1 Mar 19 15:16:20 .o0(C#++) Mar 19 15:16:25 * av500 curses his msp430 code Mar 19 15:16:31 * KotH curses with av500 Mar 19 15:16:40 av500: what's your prob? Mar 19 15:21:44 KotH: not sure Mar 19 15:21:50 adding code that does nothing does strange things Mar 19 15:22:22 in one case it inlines a function, in the other now Mar 19 15:22:23 not Mar 19 15:26:27 sounds strange Mar 19 15:26:32 what compiler? gcc? iar? Mar 19 15:26:52 gcc Mar 19 15:27:52 hm, I've just read about that gcc rewrite with c++ Mar 19 15:28:37 I just don't get it Mar 19 15:28:45 fanboiz Mar 19 15:30:26 seems liek I need to dig into it deeper, mainly to see their arguments about "c++ makes the interfaces and their boundaries more clear" Mar 19 15:31:44 I heard that the c++ rewrite will be incrementally better Mar 19 15:33:11 gcc is on it's way out Mar 19 15:33:21 unfortunately llvm is written in C++ too Mar 19 15:34:00 llvm is remarkably clean for a c++ project Mar 19 15:34:03 and for a compiler Mar 19 15:36:35 mdp: but the same could be applied to the C rewrite Mar 19 15:36:50 you didn't see what I did there Mar 19 15:37:05 nope Mar 19 15:40:11 hmm, int40_t Mar 19 15:44:16 nice Mar 19 15:44:26 when you need that extra byte Mar 19 15:45:22 does the cpu ask for 5 byte alignment too? Mar 19 15:45:29 heh Mar 19 15:50:03 "ERROR! at line 2724: [E0300] Symbol $C$PCB11 has already been defined" Mar 19 15:50:55 av500, I would not be surprised if this synopsys pcie ip wanted 5 byte alignment Mar 19 15:51:18 it's trying to dethrone musb Mar 19 15:51:41 you could build a 40-bit wide memory Mar 19 15:51:46 mdp: ;) Mar 19 15:52:06 mdp: there is always 9 bits to the byte Mar 19 15:53:15 that will work Mar 19 15:53:19 what do the french call such bytes? Mar 19 15:54:17 check with ministry of culture Mar 19 15:55:01 I'm a bit confused with i2c on the bone and DT, when I boot my bone with the 3.8 kernel I have /dev/i2c-0 and /dev/i2c-1 Mar 19 15:55:18 however in the devicetree i2c0 and i2c2 are enabled Mar 19 15:55:43 jackmitchell: presumably i2c2 is /dev/i2c-1 Mar 19 15:55:54 so does the i2c driver count how many i2c busses are enabled, rather than match up against the DT Mar 19 15:55:57 jackmitchell: sequential numbering if you don't have DT aliases Mar 19 15:56:05 jackmitchell: the N in i2c-N is just an incrementing index Mar 19 15:56:22 ok, makes sense! (in a devicetree way...) Mar 19 16:11:08 omg, all those c-fanatics here with their silly bugs are talking about c++. Mar 19 16:11:35 clash of the day Mar 19 16:11:47 jackmitchell, there might be a way to fix the order Mar 19 16:12:22 http://26-26-54.hardwarebug.org/24 Mar 19 16:13:13 panto: it's ok, I just wanted to be clear; I'm only doing some initial i2c investigations so maybe sometime in the future Mar 19 16:13:36 aholler, them are fighting words Mar 19 16:14:09 it is a pain, that on the expansion header both i2c-1 and i2c-2 are pinmuxed over the spi signals, so you can't use all the spi + clocks and i2c, unless you implemented gpio chip selects I suppose Mar 19 16:14:34 s/clocks/chip selects Mar 19 16:15:33 SPI->I2C io expander :-D Mar 19 16:16:44 SPI->I2C->I2C-Slaves+SPI? :P :D Mar 19 16:17:19 * Jacmet once did i2c-over-spi Mar 19 16:17:55 not that I'm proud of it. Mar 19 16:17:55 it's bitbanging all the way down Mar 19 16:24:40 ...to hell Mar 19 16:25:43 we once had a debug UART on SPI Mar 19 16:25:49 with some Maxim SPI to UART chip Mar 19 16:25:51 whoever split the PxSEL registers in the MSP430 needs some severe beating Mar 19 16:30:29 av500, those were pretty cool Mar 19 16:30:46 pretty useful for debugging when you couldn't spare a real uart Mar 19 16:31:03 that thing had 2 uarts, both were blocked by alternate functions Mar 19 16:31:09 luckily one of them was SPI :) Mar 19 16:31:35 yeah, most of the times you need the other functions instead of uart :) Mar 19 16:32:07 I have the debug card here somewhere, I'll hook it to the bone Mar 19 16:34:33 haha Mar 19 16:34:39 http://www.smbc-comics.com/index.php?db=comics&id=2920#comic Mar 19 16:36:54 jackmitchell: old example https://patchwork.kernel.org/patch/631941/ Mar 19 16:39:44 mdp: ah simpler than I thought; will bear that in mind, thanks Mar 19 16:41:36 jackmitchell: the driver is already bitbanging the dedicated CS line anyway, so it's a pretty straightforward replacement when run in single mode Mar 19 16:41:52 the hookup stuff for dt property etc will be different now. Mar 19 16:42:08 there's a number of other drivers that already do gpio cses so check for that Mar 19 16:42:11 for examples Mar 19 16:59:48 Vadtec: finx your connection Mar 19 16:59:56 *fix even Mar 19 17:00:12 jinx? Mar 19 17:07:08 KotH: volatile does wonders to msp430 gcc code :) Mar 19 17:07:52 is it bad to inhale volatile variables? Mar 19 17:08:20 they cause cancer in california Mar 19 17:08:29 they make my code work Mar 19 17:08:37 california causes cancer Mar 19 17:08:46 +1 Mar 19 17:10:03 hmm, this compiler has been busy on h264.c for over an hour now... Mar 19 17:10:43 Does anyone know what the "L3 memory" all the PRU docs/examples reference is? I know there is PRU data memory, DDR, but what's L3? Mar 19 17:10:44 anyone got a solution for the halting problem? Mar 19 17:10:46 it is dead jim Mar 19 17:10:58 you are using the TI compiler? Mar 19 17:11:07 cmicali: L3 cache on the processor Mar 19 17:11:10 mdp: trying to Mar 19 17:11:17 mranostay: no such thing here Mar 19 17:11:24 mru: no? Mar 19 17:11:29 L3 here refers to memory connected to the L3 interconnect Mar 19 17:11:34 av500: *g* Mar 19 17:11:36 either onchip sram or ddr Mar 19 17:11:38 mranostay: got a suggestion on the board name from the critters - Beagle-Treats Board Mar 19 17:11:53 * mranostay pukes Mar 19 17:12:00 mru: So L3 memory = PRU data memory? Mar 19 17:12:03 we are not calling it take :) Mar 19 17:12:06 cmicali: no Mar 19 17:13:28 mdp: didn't you say the ti compiler would do better than gcc? Mar 19 17:13:48 mru: better at wasting your time ;) Mar 19 17:13:57 +1 Mar 19 17:14:39 cg6x is what's actually running Mar 19 17:14:55 mru: on the beagle bone, what's connected to the L3 on the PRU? looking at the block-diagram in the ref guide it doesn't seem to specify what's connected there (i think the DDR is connected via L4, no?) Mar 19 17:15:09 ddr is on L3 Mar 19 17:15:32 as is the internal sram Mar 19 17:15:55 the L3 is the main interconnect of the chip Mar 19 17:16:18 mru: much bigger hit to get to the PRU SRAM than DDR right? :) Mar 19 17:16:42 the PRU has internal instruction and data sram Mar 19 17:17:44 mru: thanks - have not read/seen internal sram before, just seeing a lot of references in sample code and pru docs to "l3 memory" and not groking what that meant Mar 19 17:18:02 mru: yeah but the internal sram hits the l3 interconnect Mar 19 17:18:02 I assume it means any memory connected to L3 Mar 19 17:18:08 i.e. not the PRU-internal memories Mar 19 17:18:18 That Mar 19 17:18:21 that's what i thought at first Mar 19 17:18:31 mranostay: the pru-internal sram does not involve l3 when accessed from the pru Mar 19 17:18:31 according to the vague block diagram Mar 19 17:18:33 that's the point Mar 19 17:18:45 then i saw a few comments like this: This will make C30 point to 0x80000000 (L3 memory) and C31 point to 0xC0000000 (DDR memory). Mar 19 17:18:56 there's another onchip sram that's connected to l3 Mar 19 17:19:03 mru: but pru internal data sram isn't one one clock cycle right? Mar 19 17:19:10 should be Mar 19 17:19:14 from the pru Mar 19 17:19:17 not from the arm Mar 19 17:19:25 the arm accesses the pru memories over l3 Mar 19 17:19:30 when the pru is in reset Mar 19 17:19:41 mranostay: i'm not sure it is - one of the docs says even internal pru memory may not be single cycle so stick to the shared registers if you need that Mar 19 17:19:59 maybe it's not single-cycle Mar 19 17:20:03 but it's predictable Mar 19 17:20:10 i think it's not if the other PRU is accessing it Mar 19 17:20:20 L3 accesses will block unpredictably Mar 19 17:20:37 cmicali: ok that sounds right Mar 19 17:20:50 av500: gcc has become very good with optimizing unecessary stuff away. and without volatile (or explicit barriers) many reads/writes to registers the compiler would assume as unecessary just get optimized away Mar 19 17:20:53 and there is only one shared SRAM Mar 19 17:21:09 aholler: volatile is _always_ the wrong answer Mar 19 17:21:11 of course i think you address the other's SRAM if you are insane :) Mar 19 17:21:13 mru: what would you suggest the fastest way to get data off the PRU, DMA? I'm hoping to get about 1mb/s if i can. Mar 19 17:21:16 mru: no Mar 19 17:21:40 cmicali: all depends on what you're doing Mar 19 17:22:01 if you're streaming data through you should probably set up some ping-pong buffers and use dma Mar 19 17:22:27 mru: streaming data from an external A/D Mar 19 17:32:34 mru: ping pong is a fun game Mar 19 18:13:46 ping pong the new bitbang? Mar 19 18:13:56 gm mrpackethead Mar 19 18:14:10 . Mar 19 18:57:08 [kernel] koenkooi pushed 1 new commit to 3.8: http://git.io/iYPuRg Mar 19 18:57:08 kernel/3.8 a5c4357 Koen Kooi: 3.8: capebus improvements by mranostay... Mar 19 18:57:19 * mranostay blushs Mar 19 19:00:00 barmaid just crushed a glass with her bare hands Mar 19 19:00:35 or was it bear hands... Mar 19 19:03:23 sounds like a classy girl Mar 19 19:04:00 why are you ircing from a bar? Mar 19 19:04:22 because there is nothing to do Mar 19 19:11:32 mranostay: because there are no barmaids in my house Mar 19 19:13:48 well you need to change that Mar 19 19:14:22 hmm Mar 19 19:16:44 hm yes made your house a bar Mar 19 19:16:49 so you have barmaids Mar 19 19:19:36 the pub also has better beer Mar 19 19:20:28 * ogra_ wonders what mranostay finds classy about russian athlets Mar 19 19:21:21 .oO(... wimen that crush glass with a hand like a bear ... ) Mar 19 19:22:06 she does not look russian... Mar 19 19:22:14 nor sound it Mar 19 19:22:46 well, thats the immediate pic i had in my head seeing the above conversation :) Mar 19 19:26:03 bah have had worse dates :P Mar 19 19:26:13 lol Mar 19 19:28:14 that implies you had dates, so could be worse Mar 19 19:28:52 mru: that is really really low :) Mar 19 19:29:17 quality trolling ... Mar 19 19:30:04 any OSA members here? Mar 19 19:30:24 Otago Students Association Mar 19 19:30:28 not any olonger. Mar 19 19:30:31 lol Mar 19 19:30:36 optical society of america Mar 19 19:30:43 they deny me access to one of their papers Mar 19 19:30:53 thats because they don't like you Mar 19 19:31:07 everyone else can get it Mar 19 19:31:20 its a conspiracy Mar 19 19:31:38 you need to buy their glasses first Mar 19 19:31:51 +1 Mar 19 19:32:01 (dont go near the bar where mru is though) Mar 19 19:32:22 iirc KotH doesn't really drink Mar 19 19:32:52 ogra_: sure it's not about bar optics? Mar 19 19:33:12 would explain why KotH can't access Mar 19 19:33:12 hot chocolate with a russian athlete ? Mar 19 19:33:20 ah Mar 19 19:33:23 indeed Mar 19 19:33:29 ogra_: what's wrong with mru's bar? Mar 19 19:33:44 there is a bear maid crushing glasses Mar 19 19:33:46 I said she's not russian Mar 19 19:33:59 kinda hot though Mar 19 19:34:00 you said she doesnt look or sound like ... Mar 19 19:34:19 true Mar 19 19:34:19 ogra_: and what's wrong with that? Mar 19 19:34:33 dunno, but if you need the glasses ... Mar 19 19:34:37 I could find out... Mar 19 19:34:44 doit ! Mar 19 19:35:45 mru: kinda hot? that is always a cop out :) Mar 19 19:43:39 here is one for av500 Mar 19 19:43:41 http://www.nvidia.com/object/seco-dev-kit.html Mar 19 19:45:22 he's gone nvidia? Mar 19 19:46:19 that board looks horrible with that large fan Mar 19 19:46:49 aholler sure the gpu needs it Mar 19 19:46:57 is that an omap5 under that fan? Mar 19 19:47:12 russ no the desktop cuda gpu Mar 19 19:48:00 hm 530 euros Mar 19 19:48:07 deskop? it's an arm Mar 19 19:48:10 steep Mar 19 19:48:35 aholler yes arm with desktop gpu Mar 19 19:49:13 aholler, the evm he posted yesterday also has a fan Mar 19 19:49:15 woul be cool to run android and the emulate the egl stuff and than penetrating the benchmarks Mar 19 19:50:10 ogra_: haven't seen that, but I find any "embedded" board with fan just horrible ;) Mar 19 19:50:19 ++ Mar 19 19:50:39 if I would want a fan, I would guy intel Mar 19 19:50:41 though that nvidia thing is likely more for server and desktop like woglinde said Mar 19 19:50:43 hm who says that it is an embedded bord? Mar 19 19:50:53 arm goes server and desktop Mar 19 19:51:05 yup Mar 19 19:51:32 but tegra3 doesn't match what a real server needs Mar 19 19:51:49 it matches with what i.e. hadoop needs Mar 19 19:52:10 hm what they want me to say with "for professional use only" Mar 19 19:52:14 low power ... many cores and high power on demand through using the GPU Mar 19 19:52:50 you would want more than 2mb ram for hadoop I assume Mar 19 19:52:54 gb Mar 19 19:52:55 woglinde, you shouldnt build a steambox for home use with it ;) Mar 19 19:53:26 that box might be good for password cracking though Mar 19 19:53:31 calxeda doesnt have much more per node Mar 19 19:53:42 not sure what else uses cuda ;) Mar 19 19:53:56 graphics and number crunching Mar 19 19:54:09 ogra hm if valve would compile their software for arm too Mar 19 19:54:15 bitcoin mining Mar 19 19:54:21 stock exchange stuff ... Mar 19 19:54:29 ogra na Mar 19 19:54:37 you need 10 gbit ethernet for it Mar 19 19:54:40 woglinde, heh, yeah ... still trying to convince them :) Mar 19 19:54:50 for stock exchange I would assume an armv7 is too slow Mar 19 19:55:00 aholler? Mar 19 19:55:03 thats what you have the CUDA for Mar 19 19:55:08 its all about net latentcy Mar 19 19:55:17 i have started alt.bitmine Mar 19 19:55:23 but yeah, 10G is needed or a fabric Mar 19 19:55:36 cuda doesn't make i/o faster Mar 19 19:55:42 "working in a bitmine" Mar 19 19:55:56 mru: have you started stealing electricity too? Mar 19 19:56:09 currently 1ABM$ is worth $US25.00 Mar 19 19:56:17 er mrpackethead Mar 19 19:56:22 but i'll sell you some for $5.00 Mar 19 19:56:27 limited time option Mar 19 19:56:57 *but wait* theres more Mar 19 19:58:05 hmmm ... every time someone says $5 today i have too think of EA ... silly G+ Mar 19 19:58:59 EA? Mar 19 19:59:16 https://plus.google.com/109314067903163618421/posts/39kHNAzRxVx Mar 19 19:59:20 games Mar 19 19:59:22 electronic arts Mar 19 19:59:55 ah.. Mar 19 20:00:04 another thing i never have gotten into Mar 19 20:00:10 unless you count debug.exe as a game Mar 19 20:00:14 hey i worked for a EA :P Mar 19 20:00:25 what is dlc? Mar 19 20:00:34 downloadable content Mar 19 20:00:42 omg Mar 19 20:00:44 games :) Mar 19 20:00:50 gamers Mar 19 20:01:07 trolling doesn't have DLC Mar 19 20:01:12 or does it? Mar 19 20:01:24 trolling has elc Mar 19 20:01:31 heh Mar 19 20:01:33 one letter better even Mar 19 20:01:49 incrementally better ... like c++ Mar 19 20:02:31 made in mexico Mar 19 20:02:38 heh Mar 19 20:02:46 I went to EA once Mar 19 20:02:49 twas a silly place Mar 19 20:02:56 * mdp boggles at a pci express card made in mexico Mar 19 20:02:57 free2play Mar 19 20:02:58 unpossible Mar 19 20:03:05 mdp, lets see! Mar 19 20:03:11 Russ: like camelot? Mar 19 20:03:12 Russ, I don't recall you being at Embedded Alley. Mar 19 20:03:24 ea games Mar 19 20:03:29 heh Mar 19 20:03:38 panto awakes Mar 19 20:03:39 ea'ers of the world unite! Mar 19 20:03:45 I went for one day Mar 19 20:03:49 panto? Mar 19 20:03:51 I allways think those modern game copmanies are full of slaves working 80h/week Mar 19 20:03:58 yes, they are Mar 19 20:04:02 Russ, oh, you mean the uninteresting EA, gotcha ;) Mar 19 20:04:04 especially EA Mar 19 20:04:09 panto the ceo resigned because the sells werent good Mar 19 20:04:17 80h might be considered a luxury by many Mar 19 20:04:41 gee, people don't like been nickeled and dimed huh? Mar 19 20:04:43 aholler, enjoy http://ea-spouse.livejournal.com/ Mar 19 20:05:17 Russ, pics or it didn't happen? Mar 19 20:05:54 mdp, yes, show us this south of the border pci wonder Mar 19 20:09:37 done Mar 19 20:10:18 I suppose it's old enough that they hadn't yet moved manufacturing Mar 19 20:10:53 awww...how disappointing, no true wtfery Mar 19 20:11:24 it was a "hrm, that looks unusual nowadays" Mar 19 20:12:15 hey, my spudger I ordered from the cheapest ebay seller I could find was made a mere 3 miles from here Mar 19 20:13:07 crazy Mar 19 20:13:15 and I'm out Mar 19 20:13:16 bbl Mar 19 20:13:18 cya Mar 19 20:14:32 mdp: not yet moved? Mar 19 20:14:44 mdp: did usian companies use mexico as cheap labor camp? Mar 19 20:31:32 hmm, those new save/load options in menuconfig are disturbing ;) Mar 19 20:31:44 g'night Mar 19 20:54:41 * mranostay sprays prpplague with lysol Mar 19 21:20:24 * mranostay pokes kkeller Mar 19 21:20:36 ouCH! Mar 19 21:20:55 how goes it mranostay - any good trolling I missed ? Mar 19 21:30:11 mranostay: I was reading your PRU-based nixie example.. a) thank you for doing it, it is helping me :) and b) know anything about accessing DDR directly from the PRU? Mar 19 21:32:50 no but there is an example in the PRU assembler iirc Mar 19 21:33:27 i found one in the older pru_sw examples (for pruv1, not the one on beaglebone) Mar 19 21:33:46 bit confusing though - they just mmap all 256mb of ddr and start changing bytes in the example Mar 19 21:33:54 which does not seem like it would be the right way to do it Mar 19 21:34:08 heh not really the sane way to do it Mar 19 21:34:14 mmap? Mar 19 21:34:19 the pru does not have an mmu Mar 19 21:34:33 mru: sorry on the C side, it mmaps /dev/mem Mar 19 21:34:54 that's a bit evil Mar 19 21:34:59 yeah Mar 19 21:34:59 just a bit Mar 19 21:35:02 it felt quite evil to me Mar 19 21:35:35 but nobody ever achieved success by being nice Mar 19 21:35:41 what i'd like is to malloc in C and then have the PRU write to that, but unclear how i can get the physical address of that memory to give to the PRU Mar 19 21:35:47 hah Mar 19 21:35:48 evil is a necessary, hm, evil Mar 19 21:38:54 cmicali: i really don't think you can do that in userspace :) Mar 19 21:39:02 could be wrong Mar 19 21:39:16 you can mmap /dev/mem and walk the page tables :) Mar 19 21:39:34 hah Mar 19 21:39:40 yeah i did not think that was possible from userspace Mar 19 21:39:41 anyone that does that needs to be dragged out back and shot :) Mar 19 21:39:55 repeatly and then stabbed Mar 19 21:39:57 mranostay: too kind Mar 19 21:40:29 they should be tarred, feathered, stabbed, shot, hung, drawn, and quartered Mar 19 21:40:50 i guess i could make a kernel driver that does this, not sure how painful it would be to get the the pru load/uio code to work there Mar 19 21:41:13 it should be possible to use dmabuf Mar 19 21:43:29 not familiar with dmabuf but will search Mar 19 21:43:55 it's a kernel thing for sharing buffers between devices Mar 19 21:44:28 sounds like it might be good after a couple googles Mar 19 21:44:53 looks like some things use it to provide userspace access to the buffer, which would be exactly what i want Mar 19 21:45:24 that's kind of the point Mar 19 21:46:55 will investigate this, thansk Mar 19 22:38:38 * mranostay waves to pfefferz_ **** ENDING LOGGING AT Wed Mar 20 02:59:58 2013