**** BEGIN LOGGING AT Sun Nov 24 02:59:58 2019 Nov 24 04:33:56 https://youtu.be/cpqV8ubNuGU is my Penguin for now. No flight, all bite. Nov 24 04:34:05 BBBlue and ArduCopter! Nov 24 04:40:48 I think I have been discussing this issue for two or three years. I am sorry but that thing will work or I will die. Nov 24 04:40:59 Fly blue fly! Nov 24 20:20:38 hello, has anyone else has issue with the beaglebone AI basically over heating and causing issues with gcc when compling? Nov 24 20:22:09 hgfjkg: yeah the AI has some power management issues currently causing it to generate excessive heat while idle Nov 24 20:23:52 removing the DSP firmware while (or just moving/renaming it out of the way) helps, but breaks TIDL. Nov 24 20:24:10 ok thanks, i currently have my bone in a 5C room and it still is over heating L( Nov 24 20:24:12 ehh, DSP firmware file .... that was a weird typo Nov 24 20:24:59 hmm, maybe there's a way to get better convection going? like put it on its side or something Nov 24 20:25:33 hmm so maybe remove it compile opencv (which takes about 6 hours on the dual cpu) Nov 24 20:25:40 then turn it back on Nov 24 20:26:15 i have it on a lifted mesh to to keep the air moving around Nov 24 20:26:26 with software that big I'd look into the possibility of cross-compiling it from a beefy x86 machine, but maybe that's annoying with opencv due to library deps Nov 24 20:27:11 yeah the the TIDL and opencl deps cross-compiling would be a pretty big pain Nov 24 20:27:13 disabling the DSP firmware during this compilation definitely sounds like it'd be worth a try Nov 24 20:27:45 I forgot the firmware file name, check ls /lib/firmware | grep -i dra7 Nov 24 20:28:04 (and reboot after moving it out of the way) Nov 24 20:28:16 yeah I have built opencv on raspberry pi, nvidia jetson, other soc boards Nov 24 20:28:30 and each one has it's own gotcha's Nov 24 20:28:57 i wonder what fan would work to help cool it Nov 24 20:29:14 I've also made a hacky util that improves power management of the Cortex-A15 while idle, but that wouldn't be useful to you since your Cortex-A15 is the complete opposite of idle Nov 24 20:29:46 I'd say any fan would help, even just aiming a small desktop fan in the general direction of the BBAI Nov 24 20:30:30 yeah guess im off to newegg to find a 3 pin fan to hopefully help out Nov 24 20:31:16 I will play around with renaming the firmware and see how opencv reacts and detects the c66x correctly Nov 24 20:31:58 yeah the actual problem isn't the DSP itself, the problem is that the default DSP firmware brings up all four EVE cores in a way that prevents them from entering idle Nov 24 20:32:42 hmm so then opencv should still detect them like clinfo Nov 24 20:34:11 no idea, I imagine that removing the DSP firmware will may opencl support for the DSP as well, depending on how it works exactly (I have not looked into that yet) Nov 24 20:35:15 thank you very much zmatt, you got gave me path to look at and test out in a bit as the board just hit 85C and crashed Nov 24 20:35:46 anyway, I'm not really surprised it's overheating if you're stressing it at high load for hours on end without a fan Nov 24 20:36:02 that might be unavoidable even with proper power management Nov 24 20:36:28 some people have had the problem it overheats while it's sitting idle... now *that* will hpoefully be fixed with better power management :) Nov 24 20:37:01 but it's just a big powerful SoC on a tiny board Nov 24 20:37:38 yeah with using opencv the dsp, eve and cpu will be at max for hours on end when the project is finished Nov 24 20:38:35 I will let it cool down and look at better cooling for it Nov 24 20:39:12 yeah, even if better power management in the future would prevent overheating, it would have to throttle everything to do so Nov 24 20:39:12 any suggestions on a good fan? Nov 24 20:40:11 in my point of view a throttle but working is better then full speed and crashing Nov 24 20:41:24 well, throttlin is a form of not-working, at least in terms of performance targets :P Nov 24 20:41:36 a quick google reveals that https://www.element14.com/community/docs/DOC-93498/l/beaglebone-ai-frequently-asked-questions-faq includes a link to a suggested fan Nov 24 20:44:56 zmatt: You sledge hammer style PM hack seems to work okay in lieu of a fan (FWIW) Nov 24 20:45:17 ds2: yeah but probably not when you're pulling 100% cpu load Nov 24 20:45:23 like when doing a 6-hour compile job Nov 24 20:46:35 I have done 10-20 minute compiles Nov 24 20:47:20 but somehow the kernel I am using is stuck at 400MHz though Nov 24 20:47:26 which makes it dog slow compared to the jetson Nov 24 20:48:30 yeah the non-existent OPP jason added to the DT, along with lowered thermal threshold Nov 24 20:49:07 a fan would presumably help avoid this? Nov 24 20:51:35 I still think that inserting idle should be a better way to reduce power consumption (hence heat production) than lowering the cpu frequency below the lowest OPP, at least when combined with my cortex-a15 PM fix Nov 24 20:52:27 in terms of performance/power ratio Nov 24 20:52:46 I am still not convinced the fan is the best way Nov 24 20:53:00 a better heatsink or a redesigned PCB Nov 24 20:53:40 well redesigning the PCB is not exactly a very accessible option ;) and I'm not sure how much improvement there can be had on such an intrinsically cramped board Nov 24 20:53:43 the Jetson with a possible 20W dissipiation just has a giant heatsink Nov 24 20:53:46 dont know if the heatsink can be swapped out for better one without damaging the board Nov 24 20:54:03 there are crap on that board that should go Nov 24 20:54:21 getting rid of crap should free up space Nov 24 20:54:22 I'm sure there are people who will disagree with you on that Nov 24 20:54:40 those components wouldn't be on there unless someone deemed them important Nov 24 20:54:40 as a taller radiator style with a fan should keeper it cooler Nov 24 20:54:50 right btwn a choice of thermals or features, I'll take fixing thermals Nov 24 20:55:13 there are stuff there that looks like a blind dumb implementation of BBB/BBW stuff Nov 24 20:55:20 that's assuming you *can* fix the thermals Nov 24 20:55:25 that way Nov 24 20:55:54 fyi the suggested fan will not work with the bbAI as the AI requires a 3 pin not at 2pin like the older boards Nov 24 20:55:55 flipping the board over might just be enough to allow for a bigger heatsink Nov 24 20:56:07 SoC on the bottom you mean? Nov 24 20:56:16 yes so the 2 headers won't cramp things Nov 24 20:56:22 that does sound like an interesting option yes Nov 24 20:56:46 but I don't know whether or not there were good reasons to not do so Nov 24 20:57:10 alternatively, put in a mechical assembly leverage the 2 connectors as air channels and have a fan at the far end sucking air through everything (still a fan but placed differently) Nov 24 20:57:39 the heatsink right now is a hack Nov 24 20:57:53 IMO, one worse then your PM hack ;) Nov 24 20:59:23 well we can always us mineral oil hack if you want a hack to just "fix" the thermal Nov 24 21:01:13 someone should run a FEM analysis on it Nov 24 21:01:27 prehaps use this for the Soleme (sp?) stuff as a tutorial :D Nov 24 21:08:03 one thing I did notice is the PCB itself gets hot so there was some attempt in layout for the PCB handle the heat Nov 25 00:31:16 Did adafruit_bbio stop working on Buster images? Nov 25 00:32:32 I found that it just poofed away. I had it on my image and then it went gone. I could use it, write source for the LDR, and then it states that it is not on my machine. Nov 25 00:33:38 Do not fret...I will stay patient. No issue. It is not immediately necessary to answer now. Nov 25 00:33:46 I am sure everyone are doing things. Nov 25 01:19:00 Do not worry about it. I am performing a fresh image install. Nov 25 02:20:22 https://pastebin.com/aKQhY6Xa <<<< from this, I found this >>>> https://stackoverflow.com/questions/7479292/get-zerodivisionerror-float-division-in-python Nov 25 02:21:52 Nevermind. This is more of a python thing. Nov 25 02:21:57 Sorry. Nov 25 02:37:27 Hi Nov 25 02:38:32 [ 2.785847] mmcblk0: error -84 transferring data, sector 0, nr 8, cmd response 0x900, card status 0x0Loading, please wait...Gave up waiting for root device. Common problems: - Boot args (cat /proc/cmdline) - Check rootdelay= (did the system wait long enough?) - Check root= (did the system wait for the right device?) - Missing modules (cat Nov 25 02:38:33 /proc/modules; ls /dev) Nov 25 02:38:43 Im getting following error Nov 25 02:39:12 Swamy: don't try to paste multiline output in chat, it ends up a mess, use a paste service like pastebin.com Nov 25 02:39:46 Hi Zmatt Nov 25 02:39:46 it looks like it's having problems with your SD card Nov 25 02:40:03 What needs to be done Nov 25 02:40:12 use a different sd card? Nov 25 02:40:53 spl_register_fat_device: fat register err - -1 Nov 25 02:40:57 it's hard to say without more context Nov 25 02:41:08 this is output on the serial console? Nov 25 02:41:34 also, what hardware are we talking about? Nov 25 02:41:51 Yes when board is booting SDCard console is getting following msg Nov 25 02:42:20 can you just share the full output? (e.g. using pastebin.com) Nov 25 02:42:22 Im using Beaglebone Nov 25 02:44:27 U-Boot SPL 2014.04-00014-g47880f5 (Apr 22 2014 - 13:23:54)** Partition 1 not valid on device 0 **spl_register_fat_device: fat register err - -1### ERROR ### Please RESET the board ### Nov 25 02:46:03 https://pastebin.com/G7AHm7eD Nov 25 02:47:13 Swamy: seems like your sd card simply died Nov 25 02:47:50 (you were also running quite an old system it seems, but that aside) Nov 25 02:48:08 Okay I will Check with other sdcard Nov 25 02:48:48 Swamy: in general, any type of NAND flash based storage (including SD cards and eMMC) has limited lifetime, dependent on often data is being written to it Nov 25 02:49:50 consumer SD cards having the shortest lifetime of all Nov 25 02:49:54 https://pastebin.com/gmEcjG5b Nov 25 02:50:29 and one more issue booting with prebuilt images Nov 25 02:50:40 from EMMC flash Nov 25 02:51:52 really hard to say what's going on here, what image are you using? Nov 25 02:52:17 looks like it's a very old one Nov 25 02:52:26 consider reflashing to the latest image Nov 25 02:53:52 Okay i will try and update with the latest images Nov 25 02:54:39 looks like there's some sort of device tree problems with your current system but other than that it's hard to tell from this output **** ENDING LOGGING AT Mon Nov 25 02:59:58 2019