**** BEGIN LOGGING AT Wed Feb 12 03:00:34 2020 Feb 12 08:14:31 Morning! Feb 12 08:22:30 morning Feb 12 08:41:37 @Tofe @Herrie Hi! Feb 12 08:42:15 @Tofe @Herrie Allow me to repeat yesterday's questions, please be so kind to help me one more time! Feb 12 08:43:06 @Tofe @Herrie With formatting data is meant to format data the way TWRP asks you to type 'YES', is that right? Feb 12 08:43:41 @Tofe @Herrie 2.nd LuneOs starts from slot A, did I remember that correctly? Feb 12 08:44:45 @Tofe @Herrie 3.rd Do I have to check the checkbox:'Wipe Dalvik' in the ADB sideload dialogue in twrp when flashing LuneOS? Thank you! Feb 12 08:45:49 19:30 <+Tofe> windknecht: yes, yes, no :) Feb 12 08:46:07 if you have used the same order of questions, then here ^ is your answer from yesterday Feb 12 08:46:57 windknecht: Wipe Dalvik didn't really matter much in my case, but well Feb 12 08:47:46 @JaMa @Tofe @Herrie Thank you! I'll try one last time! Feb 12 08:48:11 I really apreciate your helpfulness! Feb 12 08:48:28 Herrie: any luck with the benchmark? Feb 12 08:48:44 JaMa: Seems it locked up my machine after many hours Feb 12 08:48:50 SO I guess I have some partial logs Feb 12 08:49:28 that's good, if it finished some of the steps completely you can still restart the remaining steps (by copy-pasting their section from test.sh) Feb 12 08:53:06 Seems 1-4 completed, 5 failed Feb 12 08:53:10 I'll retry from 5-7 Feb 12 08:54:25 4 is the worst (too many bitbake threads) Feb 12 08:55:08 3 tasks failed with 4 for me Feb 12 08:55:43 Top20 of 4: https://paste.ubuntu.com/p/dsMHTPs34F/ Feb 12 08:58:37 all 3 ICEs? Feb 12 09:00:26 and it was qtbase right, so it removed qtwebengine and qtwebkit from your top20 Feb 12 09:00:58 and looks like firefox or one of its dependency failed as well Feb 12 09:01:34 "/home/herriemerim/LuneOS/oetest/test-oe-build-time/poky/build/tmp/work/core2-64-poky-linux/coreutils/8.31-r0/recipe-sysroot/usr/include/sys/types.h:164:1: internal compiler error: Segmentation fault" Feb 12 09:02:13 how hot was it running? it's really demanding, but in worst case it should throttle, before triggering ICEs and you would be probably seeing ICEs in LuneOS builds as well Feb 12 09:02:56 even my temporary 1600AF was able to survive it in the end (with a lot of swap) Feb 12 09:03:16 "error: Could not compile `rustc`." Feb 12 09:03:31 "(signal: 11, SIGSEGV: invalid memory reference)" Feb 12 09:03:53 do you have it overclocked? Feb 12 09:04:08 or with some aggressive XMP? Feb 12 09:04:09 And qtbase: /home/herriemerim/LuneOS/oetest/test-oe-build-time/poky/build/tmp/work/core2-64-poky-linux/qtbase/5.13.2+gitAUTOINC+a7a24784ee-r0/git/src/gui/painting/qpainterpath.cpp:3607:1: internal compiler error: Segmentation fault Feb 12 09:04:17 JaMa: No, all standard Feb 12 09:04:24 I think it could be just running too hot Feb 12 09:04:38 and have you run memtest86 ever on this? Feb 12 09:04:39 I like the quietness of the Noctua cooler, but it's cooling capacities isn't great Feb 12 09:04:51 memtest86 I can check. Don't think I ran it Feb 12 09:05:50 then please run just build 7 if you can remotely and then run memtest86 over night Feb 12 09:06:20 you can still RMA those modules? Feb 12 09:07:26 Well I'm not sure if it are the modules. I think CPU is getting to it's max allowed temperature really Feb 12 09:07:33 And that causes issues. Feb 12 09:07:38 I have this one on it: https://noctua.at/en/nh-u14s-tr4-sp3 Feb 12 09:07:59 that should cause throttling, not segfaults Feb 12 09:08:31 Running the other builds now and will then do memtest86 Feb 12 09:08:39 memory modules getting too hot, because of close to hot cpu might be a factor too Feb 12 09:08:46 Memory has 2 yr warrantee so can always returns them Feb 12 09:09:03 What I understood Threadripper also has some "bugs" that could cause this Feb 12 09:09:59 btw: to run memtest86 you need to switch booting to legacy in bios (if you're normally using uefi boot), I was wondering why I don't see memtest86 in the grub boot menu for a while.. Feb 12 09:10:43 https://ibb.co/xHryQjR Feb 12 09:10:47 That's my current CPU temp Feb 12 09:11:51 well it has architecture disadvantage for only 2 dies from 4 having the direct memory access and 4 NUMA domains, while 3rd generation has shared memory channels in central IO die Feb 12 09:12:10 but I'm not aware of any bugs which should cause this Feb 12 09:12:14 Max can be 68c according to AMD, so it's not there yet, but it gets close at some points Feb 12 09:13:27 The cooler should be OK, it's not the best performing one, but decent according to: https://www.phoronix.com/scan.php?page=article&item=amd-2990wx-cooling&num=3 Feb 12 09:13:53 agreed Feb 12 09:14:35 do you see some frequency scalling during the build? Feb 12 09:14:44 Not sure how to check that? Feb 12 09:15:04 Never really looked into it Feb 12 09:15:42 watch -n 2 grep MHz /proc/cpuinfo Feb 12 09:16:28 it should oscilate between 3GHz and 4.2 and drop lower once it gets too hot Feb 12 09:16:40 It seems to drop a little here and there Feb 12 09:17:00 well it also drops a lot when some core is really idle Feb 12 09:17:01 Mostly 3200 and 3300 Mhz and sometimes I see a few 2600 and 2800 Feb 12 09:17:40 I'll log it to a file for a few mins Feb 12 09:17:55 But it mainly is 3,2/3,3 GHz Feb 12 09:18:01 that should be fine Feb 12 09:18:02 From all items on screen, there are 2-3 below that Feb 12 09:18:21 it wont boost all the way to 4+ GHz when it's too hot for TDP Feb 12 09:18:39 but that's expected with all-core load from OE build Feb 12 09:20:50 CPU seems solid, doesn't drop anywhere serious I'd say Feb 12 09:21:03 I didn't play much with settings of memory in bios etc Feb 12 09:21:26 do you remember what sticks you have and what frequency they are running at? Feb 12 09:22:26 I might want to update BIOS though Feb 12 09:22:28 well I can overload the cooling in beefy server with the inlet air from AC on reasonable 17C :) Feb 12 09:22:44 http://www.asrock.com/mb/AMD/X399%20Taichi/index.asp#BIOS Feb 12 09:22:47 yeah, might help as well Feb 12 09:22:48 I think I'm still on 3.50 Feb 12 09:23:30 Yup on 3.50 Feb 12 09:24:51 These modules: https://www.amazon.de/gp/product/B077HYB2JP/ref=ppx_yo_dt_b_asin_title_o01_s00?ie=UTF8&psc=1 Feb 12 09:25:32 I couldn't care less about the RGB, but it were the only decent modules that were compatible for a decent price Feb 12 09:25:55 wow fancy 3800Mhz, I guess it runs slower due to threadripper Feb 12 09:26:12 Yeah Feb 12 09:26:42 the sweet sport for ryzen 3 is 3600, I think for 2nd gen is a bit lower Feb 12 09:26:50 spot Feb 12 09:27:37 https://paste.ubuntu.com/p/zJNW7gkxcF/ Feb 12 09:27:59 2666 Mhz Feb 12 09:28:10 Yeah I could probably push that a bit if needed but well Feb 12 09:28:40 2933 is max (without OC) for 2nd gen (but probably not with all 8 sticks) Feb 12 09:29:18 but especially as it's running on 2666 it should be rock solid Feb 12 09:29:24 It should be Feb 12 09:29:25 lets see what memtest says Feb 12 09:30:03 I'll also update BIOS, that usually doesn't hurt Feb 12 09:30:31 yes the agesa update might help Feb 12 09:30:45 it might be just the stress of memory controller with all 8 slots being used Feb 12 09:31:28 It seems the modules should be able to do 2933, but having 8*16GB limits options in general Feb 12 09:31:36 Compared to 2 or 4 modules ;) Feb 12 09:34:45 right Feb 12 09:36:33 on 3rd gen it's 3200 with 4 sticks, 2933 with 8x8, 2667 with 8x16 1R or 8x32 2R Feb 12 09:37:12 your are 8x16 but 2R, so it might be quite heavy on the controller Feb 12 09:38:44 Yeah anyway I'm pretty happy with the performance, but would be good if more stable Feb 12 10:00:54 yeah your chromium-x11 time is much better than the 2 servers I was testing, but for you it was running without qtweb* build in parallel, so lets see when there are more complete results Feb 12 10:21:56 I've added some memory bandwidth test results https://github.com/shr-project/test-oe-build-time/commit/2ff2d53861fa64e8228de9d8aec548b9c3b491b8 Feb 12 10:34:47 Herrie: your amazon link doesn't seem accurate, based on your pastebin you have these 2933Mhz https://www.amazon.com.au/Corsair-CMR128GX4M8Z2933C16-Vengeance-PC4-23400-Desktop/dp/B077HYCZ4Q not 3800MHz which makes more sense as it wont run faster than 2933 anyway Feb 12 11:11:48 JaMa, Well I got the link from my Amazon history, could be they link to a different product Feb 12 11:17:01 But yeah that makes more sens Feb 12 11:17:32 Also seeing that CMR128GX4M8Z2933C16 is on the memory QVL of my Asrock X399 Taichi Feb 12 11:17:42 I bought it based on QVL just to be sure it would work Feb 12 11:31:47 JaMa: The 5 is almost done running 7759/7999 it seems Feb 12 11:41:42 and your dmidecode shows "Part Number: CMR128GX4M8Z2933C16" Feb 12 11:46:15 JaMa: Yes, seems Amazon just linked to a different product with my order history :S Feb 12 12:07:58 Herrie: as long as you paid for the right cheaper set :) Feb 12 12:29:45 * Tofe is looking at his rig and sees an old rock with a screen Feb 12 13:18:53 JaMa: Top 20 for the 8 threads: https://paste.ubuntu.com/p/s5Nj8rQcR4/ Feb 12 13:19:29 8 errors though Feb 12 13:19:48 Ah seems i'm running out of space :P Feb 12 13:20:33 Let me throw away some Halium stuff Feb 12 13:22:50 Top 20 for 16 threads: https://paste.ubuntu.com/p/HKD9kf4kgN/ Feb 12 13:24:52 looks like qtbase finished this time, but chromium didn't :) Feb 12 13:25:16 any errors in 16 threads? Feb 12 13:25:39 Nope, seems OK: NOTE: Tasks Summary: Attempted 7999 tasks of which 1 didn't need to be rerun and all succeeded. Feb 12 13:25:39 66.19user 10.38system 2:31:26elapsed 0%CPU (0avgtext+0avgdata 29584maxresident)k Feb 12 13:25:39 11576inputs+5408outputs (14major+64962minor)pagefaults 0swaps Feb 12 13:26:23 pretty good compared to https://github.com/shr-project/test-oe-build-time/blob/master/dual-Xeon-E5-2699-v3-8-channels/6-build-16-bb-threads.top20 https://github.com/shr-project/test-oe-build-time/blob/master/dual-Xeon-E5-2699-v3-8-channels/results.txt Feb 12 13:26:40 actually yours is 9 mins faster in total Feb 12 13:30:09 Could be my disks? Feb 12 13:30:15 Connection doesn't matter here right? Feb 12 13:31:42 Value for money for sure Threadripper is good ;) Feb 12 13:32:32 Seeing that the CPU's alone for that builder are like EUR 9000,- ;) Feb 12 13:35:47 I got 1Gbit/1Gbit connection now (they upgraded speed for lower monthly payment recently :D) and M.2 SSD ;) Feb 12 13:36:11 The m.2 is probably quicker v.s. the RAID0 Feb 12 13:36:53 do you have sata or nvme drive? Feb 12 13:39:30 my tmpfs build on 8 channel MC is finishing now, will collect all results in one spreadsheet later to make it more easy to compare Feb 12 13:39:59 will you be able to send me the full results some day later? Feb 12 13:41:46 nvme ;) Feb 12 13:41:53 1TB, might at some point upgrade to 2TB Feb 12 13:42:08 JaMa: Yeah will re-run soonish I guess Feb 12 16:08:09 JaMa: The 2 bb threads is not very quick.. Also CPU speed regularly drops to 1.x Ghz there Feb 12 16:11:49 yes that's expected **** ENDING LOGGING AT Thu Feb 13 02:59:57 2020