Message boards :
Graphics cards (GPUs) :
Not getting new work for some graphics cards.
Message board moderation
| Author | Message |
|---|---|
|
Send message Joined: 16 May 13 Posts: 41 Credit: 145,731,947 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
One host with an Nvidia GT 710 is getting work for cpu and gpu. My other host with two Nvidia GTX 670 is only getting work for cpu and not for gpu - the settings are the same. Why is one host not getting new work for gpu? |
|
Send message Joined: 13 Dec 17 Posts: 1419 Credit: 9,119,446,190 RAC: 891 Level ![]() Scientific publications ![]() ![]() ![]() ![]()
|
Hope the app developer jumps in. Only thing I can think of is a Compute Capability version limitation in the app. What do the first 30 lines in the Event Log show after initial BOINC startup? BOINC prints out what it finds to compute with and any error messages tell the tale. |
|
Send message Joined: 16 May 13 Posts: 41 Credit: 145,731,947 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
Sa 20 Jan 2018 10:16:07 CET | | Starting BOINC client version 7.6.31 for x86_64-pc-linux-gnu Sa 20 Jan 2018 10:16:07 CET | | log flags: file_xfer, sched_ops, task Sa 20 Jan 2018 10:16:07 CET | | Libraries: libcurl/7.47.0 OpenSSL/1.0.2g zlib/1.2.8 libidn/1.32 librtmp/2.3 Sa 20 Jan 2018 10:16:07 CET | | Data directory: /var/lib/boinc-client Sa 20 Jan 2018 10:16:07 CET | | CUDA: NVIDIA GPU 0: GeForce GTX 670 (driver version 384.11, CUDA version 9.0, compute capability 3.0, 1996MB, 1863MB available, 2634 GFLOPS peak) Sa 20 Jan 2018 10:16:07 CET | | CUDA: NVIDIA GPU 1: GeForce GTX 670 (driver version 384.11, CUDA version 9.0, compute capability 3.0, 1999MB, 1955MB available, 2634 GFLOPS peak) Sa 20 Jan 2018 10:16:07 CET | | OpenCL: NVIDIA GPU 0: GeForce GTX 670 (driver version 384.111, device version OpenCL 1.2 CUDA, 1996MB, 1863MB available, 2634 GFLOPS peak) Sa 20 Jan 2018 10:16:07 CET | | OpenCL: NVIDIA GPU 1: GeForce GTX 670 (driver version 384.111, device version OpenCL 1.2 CUDA, 1999MB, 1955MB available, 2634 GFLOPS peak) Sa 20 Jan 2018 10:16:08 CET | | Host name: alien Sa 20 Jan 2018 10:16:08 CET | | Processor: 6 AuthenticAMD AMD FX(tm)-6100 Six-Core Processor [Family 21 Model 1 Stepping 2] Sa 20 Jan 2018 10:16:08 CET | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf pni pclmulqdq monitor ssse3 cx16 sse4_1 sse4_2 popcnt aes xsave avx lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs xop skinit wdt lwp fma4 nodeid_msr topoext perfctr_core perfctr_nb cpb hw_pstate vmmcall arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold Sa 20 Jan 2018 10:16:08 CET | | OS: Linux: 4.13.0-26-generic Sa 20 Jan 2018 10:16:08 CET | | Memory: 15.69 GB physical, 15.97 GB virtual Sa 20 Jan 2018 10:16:08 CET | | Disk: 212.90 GB total, 134.00 GB free Sa 20 Jan 2018 10:16:08 CET | | Local time is UTC +1 hours Sa 20 Jan 2018 10:16:08 CET | | Config: GUI RPCs allowed from: Sa 20 Jan 2018 10:16:08 CET | Collatz Conjecture | URL https://boinc.thesonntags.com/collatz/; Computer ID 826075; resource share 400 Sa 20 Jan 2018 10:16:08 CET | GPUGRID | URL http://www.gpugrid.net/; Computer ID 452656; resource share 600 Sa 20 Jan 2018 10:16:08 CET | PrimeGrid | URL http://www.primegrid.com/; Computer ID 907161; resource share 100 Sa 20 Jan 2018 10:16:08 CET | Rosetta@home | URL http://boinc.bakerlab.org/rosetta/; Computer ID 3332621; resource share 100 Sa 20 Jan 2018 10:16:08 CET | GPUGRID | General prefs: from GPUGRID (last modified 17-Dec-2017 16:39:10) Sa 20 Jan 2018 10:16:08 CET | GPUGRID | Host location: none Sa 20 Jan 2018 10:16:08 CET | GPUGRID | General prefs: using your defaults Sa 20 Jan 2018 10:16:08 CET | | Reading preferences override file Sa 20 Jan 2018 10:16:08 CET | | Preferences: Sa 20 Jan 2018 10:16:08 CET | | max memory usage when active: 8035.34MB Sa 20 Jan 2018 10:16:08 CET | | max memory usage when idle: 14463.62MB Sa 20 Jan 2018 10:16:12 CET | | max disk usage: 137.95GB Sa 20 Jan 2018 10:16:12 CET | | max CPUs used: 2 Sa 20 Jan 2018 10:16:12 CET | | (to change preferences, visit a project web site or select Preferences in the Manager) Sa 20 Jan 2018 10:16:12 CET | GPUGRID | Sending scheduler request: To fetch work. Sa 20 Jan 2018 10:16:12 CET | GPUGRID | Requesting new tasks for CPU and NVIDIA GPU Sa 20 Jan 2018 10:16:16 CET | GPUGRID | Scheduler request completed: got 0 new tasks Sa 20 Jan 2018 10:16:16 CET | GPUGRID | No tasks sent |
|
Send message Joined: 13 Dec 17 Posts: 1419 Credit: 9,119,446,190 RAC: 891 Level ![]() Scientific publications ![]() ![]() ![]() ![]()
|
Well, there is not a clue in the startup messages. Both cards are seen and everything looks normal. No error messages. The usual thing to look at when there a more than one card in a system of differing compute capabilities is to set the <use_all_gpus>1</use_all_gpus> flag in cc_config.xml. Boinc normally only uses the card with the highest compute capability but both of your cards are seen and are ready for use. So that is not the case. So, back to my original premise that the app has a restriction on the level of compute capability allowed. |
|
Send message Joined: 16 May 13 Posts: 41 Credit: 145,731,947 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
I think I should really search for another project. Since I've been here, there are always bugs - some computer get work, some not. No other boinc projekt has this problems. Mostly you don't get any reply. So I will crunch to some projects which needs my computer. |
|
Send message Joined: 5 Jan 09 Posts: 670 Credit: 2,498,095,550 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I think I should really search for another project. It probably is to do with the fact that GTX670 is a CC3.0 card and can't run the GpuGrid Cuda 8.0 App but can run the the Cuda 6.5 App. In which case you will need to downgrade your host's drivers to a pre Cuda 8.0 driver. |
|
Send message Joined: 16 May 13 Posts: 41 Credit: 145,731,947 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
The new WUs are not supported by the old driver. |
|
Send message Joined: 5 Jan 09 Posts: 670 Credit: 2,498,095,550 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
yes, they are on Windows at least not sure about linux. |
|
Send message Joined: 16 May 13 Posts: 41 Credit: 145,731,947 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
Not on linux. I tested it. |
©2025 Universitat Pompeu Fabra