CUDA GPU workunits -- processing errors

Questions and Answers : GPU applications : CUDA GPU workunits -- processing errors

Author	Message
The Weasel Send message Joined: 6 Jun 99 Posts: 126 Credit: 46,539,432 RAC: 22,206	Message 1278503 - Posted: 31 Aug 2012, 23:24:14 UTC Last modified: 1 Sep 2012, 0:19:12 UTC
	I thought it was just my machines, but it seems more people are having this problem. Usually when an update comes out I download it for all of my systems. Any of my systems without any Nvidia cards are running fine. But anything with a card has a minimum of 10-15 errors. I have multiple different cards on different systems. One older system in particular has never had any errors and now has 11 errors, it is an old system with a 8600GTS card, but was flawless until all of the updates. Has anyone found the most stable version of seti/Nvidia drivers that do not have so many errors? ____________
	ID: 1278503 ·

Gatekeeper Send message Joined: 14 Jul 04 Posts: 757 Credit: 144,178,374 RAC: 120,821	Message 1278675 - Posted: 1 Sep 2012, 5:41:03 UTC - in response to Message 1278503. Last modified: 1 Sep 2012, 5:50:10 UTC
	I thought it was just my machines, but it seems more people are having this problem. Usually when an update comes out I download it for all of my systems. Any of my systems without any Nvidia cards are running fine. But anything with a card has a minimum of 10-15 errors. I have multiple different cards on different systems. One older system in particular has never had any errors and now has 11 errors, it is an old system with a 8600GTS card, but was flawless until all of the updates. Has anyone found the most stable version of seti/Nvidia drivers that do not have so many errors? I assume you're using Lunatics opti-apps, and you appear to have 301.42 across the board. That's a good stable combination for Fermi cards, and SHOULD work OK with the older cards as well. EDIT: You've got errors on both CPU and GPU work. There are alot of different codes, so my guess is that it's not related to NVidia drivers or Lunatics. The only one I can speak of with certainty is the error code (-12),and that is a normal error that you will encounter from time to time on GPU work. ____________
	ID: 1278675 ·

Ageless Send message Joined: 9 Jun 99 Posts: 11761 Credit: 2,075,606 RAC: 2,429	Message 1278689 - Posted: 1 Sep 2012, 6:03:16 UTC - in response to Message 1278503.
	Usually when an update comes out I download it for all of my systems. Why? Most of what's in those updates is bug-fixes for games. Are you playing a lot of (newer) games on all your systems? It's really not necessary to update your drivers ad nauseam, just to have them 'up-to-date'. When your work runs faultless using a certain driver, why the need to update? What could a newer driver do to better that performance? ____________ Jord - BOINC FAQ Service - BOINC User Wiki Real is just a matter of perception.
	ID: 1278689 ·

The Weasel Send message Joined: 6 Jun 99 Posts: 126 Credit: 46,539,432 RAC: 22,206	Message 1278800 - Posted: 1 Sep 2012, 11:16:06 UTC - in response to Message 1278675.
	Yes, I have errors for both GPU and CPU, but only on systems that have a GPU. The other systems that have only a CPU seem to not have any errors. I have not played any games for about the last month and if I do play I turn of SETI first and restart after I am done. I have not installed optimised after the latest updates have come out unless it is still seeing the old optimised? ____________
	ID: 1278800 ·

BilBg Send message Joined: 27 May 07 Posts: 2089 Credit: 3,270,621 RAC: 8,143	Message 1278814 - Posted: 1 Sep 2012, 12:00:59 UTC - in response to Message 1278800.
	I have not installed optimised after the latest updates have come out unless it is still seeing the old optimised? What 'latest updates' are you talking about ('latest' updates for Windows? for BOINC? for NVIDIA? ...) No update (even for BOINC) will change your 'optimised' (SETI apps), Only new Lunatics Installer or detach from SETI ([Remove] SETI) can change/delete them. ____________ - ALF - "Find out what you don't do well ..... then don't do it!" :)
	ID: 1278814 ·

The Weasel Send message Joined: 6 Jun 99 Posts: 126 Credit: 46,539,432 RAC: 22,206	Message 1279315 - Posted: 2 Sep 2012, 17:26:57 UTC
	I just got back from out of town, I will look at what I have installed, I had updated BOINC version and Nvidia drivers, I thought they would overwrite any of the previous Lunatics installation. I wanted to get them running stock with no errors before I install Lunatics. I will go through the computers later tonight and reinstall. Thanks. ____________
	ID: 1279315 ·

johndad5 Send message Joined: 1 Feb 03 Posts: 1 Credit: 1,644,697 RAC: 0	Message 1298398 - Posted: 24 Oct 2012, 12:44:53 UTC
	It appears my NVidia card is consistently erroring. I have the following configuration followed by the error. Please help! 10/21/2012 9:40:21 AM \| \| No config file found - using defaults 10/21/2012 9:40:23 AM \| \| Starting BOINC client version 7.0.28 for windows_x86_64 10/21/2012 9:40:23 AM \| \| log flags: file_xfer, sched_ops, task 10/21/2012 9:40:23 AM \| \| Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6 10/21/2012 9:40:23 AM \| \| Data directory: C:\ProgramData\BOINC 10/21/2012 9:40:23 AM \| \| Running under account john 10/21/2012 9:40:23 AM \| \| Processor: 8 GenuineIntel Intel(R) Core(TM) i7-3610QM CPU @ 2.30GHz [Family 6 Model 58 Stepping 9] 10/21/2012 9:40:23 AM \| \| Processor: 256.00 KB cache 10/21/2012 9:40:23 AM \| \| Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 syscall nx lm vmx tm2 popcnt aes pbe 10/21/2012 9:40:23 AM \| \| OS: Microsoft Windows 7: Home Premium x64 Edition, Service Pack 1, (06.01.7601.00) 10/21/2012 9:40:23 AM \| \| Memory: 5.90 GB physical, 11.81 GB virtual 10/21/2012 9:40:23 AM \| \| Disk: 653.44 GB total, 551.94 GB free 10/21/2012 9:40:23 AM \| \| Local time is UTC -4 hours 10/21/2012 9:40:23 AM \| \| NVIDIA GPU 0: GeForce GTX 660M (driver version 306.97, CUDA version 5.0, compute capability 3.0, 2048MB, 8382398MB available, 730 GFLOPS peak) 10/21/2012 9:40:23 AM \| \| OpenCL: NVIDIA GPU 0: GeForce GTX 660M (driver version 306.97, device version OpenCL 1.1 CUDA, 2048MB, 8382398MB available) 10/21/2012 9:40:23 AM \| rosetta@home \| URL http://boinc.bakerlab.org/rosetta/; Computer ID 1566392; resource share 80 10/21/2012 9:40:23 AM \| SETI@home \| URL http://setiathome.berkeley.edu/; Computer ID 6778928; resource share 80 10/21/2012 9:40:23 AM \| WUProp@Home \| URL http://wuprop.boinc-af.org/; Computer ID 41128; resource share 15 10/21/2012 9:40:23 AM \| WUProp@Home \| General prefs: from WUProp@Home (last modified 01-Oct-2012 08:42:43) 10/21/2012 9:40:23 AM \| WUProp@Home \| Host location: none 10/21/2012 9:40:23 AM \| WUProp@Home \| General prefs: using your defaults 10/21/2012 9:40:23 AM \| \| Preferences: 10/21/2012 9:40:23 AM \| \| max memory usage when active: 604.64MB 10/21/2012 9:40:23 AM \| \| max memory usage when idle: 4232.45MB 10/21/2012 9:40:23 AM \| \| max disk usage: 10.00GB 10/21/2012 9:40:23 AM \| \| don't compute while active 10/21/2012 9:40:23 AM \| \| don't use GPU while active 10/21/2012 9:40:23 AM \| \| suspend work if non-BOINC CPU load exceeds 25 % 10/21/2012 9:40:23 AM \| \| (to change preferences, visit the web site of an attached project, or select Preferences in the Manager) 10/21/2012 9:40:23 AM \| \| Not using a proxy Stderr output <core_client_version>7.0.28</core_client_version> <![CDATA[ <message> Incorrect function. (0x1) - exit code 1 (0x1) </message> <stderr_txt> setiathome_CUDA: Found 1 CUDA device(s): Device 1 : GeForce GTX 660M totalGlobalMem = -2147483648 sharedMemPerBlock = 49152 regsPerBlock = 65536 warpSize = 32 memPitch = 2147483647 maxThreadsPerBlock = 1024 clockRate = 950000 totalConstMem = 65536 major = 3 minor = 0 textureAlignment = 512 deviceOverlap = 1 multiProcessorCount = 2 setiathome_CUDA: CUDA Device 1 specified, checking... Device 1: GeForce GTX 660M is okay SETI@home using CUDA accelerated device GeForce GTX 660M setiathome_enhanced 6.09 Visual Studio/Microsoft C++ libboinc: 6.3.22 Work Unit Info: ............... WU true angle range is : 0.361654 Optimal function choices: ----------------------------------------------------- name ----------------------------------------------------- v_BaseLineSmooth (no other) v_GetPowerSpectrum 0.00024 0.00000 v_ChirpData 0.01323 0.00000 v_Transpose4 0.00250 0.00000 FPU opt folding 0.00238 0.00000 CUFFT error in file 'd:/Projects/SETI/seti_boinc/client/cuda/cudaAcc_fft.cu' in line 62. </stderr_txt> ]]> ____________
	ID: 1298398 ·

Colin Send message Joined: 21 Feb 10 Posts: 12 Credit: 1,493,271 RAC: 0	Message 1298461 - Posted: 25 Oct 2012, 0:07:20 UTC
	I also have been told that my NVidia 560 Ti errors are adversely affecting the system. Since I don't know how to stop receiving work for my GPU I guess I will have to bail out of SETI. Colin ____________
	ID: 1298461 ·

Volunteer tester Send message Joined: 9 Apr 02 Posts: 12012 Credit: 19,591,637 RAC: 43,556	Message 1298469 - Posted: 25 Oct 2012, 1:16:29 UTC - in response to Message 1298461. Last modified: 25 Oct 2012, 1:44:49 UTC
	Have you tried checking your online preferences? Personally, though, I'd try to fix the problems as the GPU can outproduce the CPU any time. My friend's 560 Ti is an amazing value for crunching.
	ID: 1298469 ·

Ageless Send message Joined: 9 Jun 99 Posts: 11761 Credit: 2,075,606 RAC: 2,429	Message 1298561 - Posted: 25 Oct 2012, 7:52:14 UTC - in response to Message 1298461.
	Since I don't know how to stop receiving work for my GPU I guess I will have to bail out of SETI. What's wrong with asking for help? project preferences (they are part of Your account). Edit them. Uncheck "Use NVIDIA GPU" Uncheck "If no work for selected applications is available, accept work from other applications?" Save preferences through the "update preferences" button. Open BOINC Manager. Advanced view. Projects tab. Select Seti. Click Update. From that time forward BOINC will only request CPU work from Seti. As for your actual problems, if you want to fix them, you could try to update your videocard drivers. Was 266.44 even a driver geared for the 5xx range of Nvidia? I thought it needed 275 and up to work correctly. ____________ Jord - BOINC FAQ Service - BOINC User Wiki Real is just a matter of perception.
	ID: 1298561 ·

Previous · 1 · 2

Questions and Answers : GPU applications : CUDA GPU workunits -- processing errors