Message boards :
News :
GPU Issues Mega Thread
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 8 · Next
Author | Message |
---|---|
Send message Joined: 28 Sep 13 Posts: 4 Credit: 6,500,579 RAC: 0 |
@Jake Weiss, I think there is something wrong with you identification or calculation for the estimated GFLOP/s of my grafic cards. How you can see in the log file of my WU's. http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1761924468 Estimated Nvidia GPU GFLOP/s: 1066 SP GFLOP/s, 133 DP FLOP/s But this cards should have more power, something similar to this data 5120 SP GFLOP/s, 1706 DP FLOP/s. Tim |
Send message Joined: 8 Apr 09 Posts: 70 Credit: 11,027,167,827 RAC: 0 |
Hello Tim, Yes, you are right the Numbers are wrong. The numbers of my AMD cards are there where they should be. I will test MW on my Titan blacks tomorrow. Did you enable inside the driver the double precision option? If this is disabled, then the double precison number could be true. And even when you get the Titan Blacks fully utilized, then it is slower than a AMD card with one Terraflop. Nvidia does not really care about OpenCL, they still pushing CUDA. So OpenCL is not optimized very well. |
Send message Joined: 6 Apr 12 Posts: 42 Credit: 3,215,609 RAC: 0 |
Just to note my NVIDIA GT730 (that has no fan) works great. |
Send message Joined: 10 Feb 09 Posts: 52 Credit: 16,291,993 RAC: 0 |
I know exactly what you mean about the programming minefield that is OpenCL. The great power of OpenCl: runs everywhere (cpu, gpu, fpga, Arm, whatsoever...) The great weakness of OpenCl: runs everywhere (cpu, gpu, fpga, Arm, whatsoever...) |
Send message Joined: 28 Sep 13 Posts: 4 Credit: 6,500,579 RAC: 0 |
Hello Tim, Double Precision is activated. |
Send message Joined: 31 Aug 16 Posts: 2 Credit: 1,503,609 RAC: 0 |
in my case I think the problem is it use the wrong GPU, intel instead of the nvidia: Using SSE4.1 path Found 1 platform Platform 0 information: Name: Apple Version: OpenCL 1.2 (Jun 30 2016 20:18:53) Vendor: Apple Extensions: cl_APPLE_SetMemObjectDestructor cl_APPLE_ContextLoggingFunctions cl_APPLE_clut cl_APPLE_query_kernel_names cl_APPLE_gl_sharing cl_khr_gl_event Profile: FULL_PROFILE Didn't find preferred platform Using device 0 on platform 0 Found 2 CL devices Device 'Iris Pro' (Intel:0x1024500) (CL_DEVICE_TYPE_GPU) Board: Driver version: 1.2(Aug 29 2016 22:20:39) Version: OpenCL 1.2 Compute capability: 0.0 Max compute units: 40 Clock frequency: 1200 Mhz Global mem size: 1610612736 Local mem size: 65536 Max const buf size: 65536 Double extension: (none) Device doesn't support double precision Failed to calculate likelihood <background_integral> nan </background_integral> <stream_integral> nan nan nan </stream_integral> <background_likelihood> nan </background_likelihood> <stream_only_likelihood> nan nan nan </stream_only_likelihood> <search_likelihood> nan </search_likelihood> 00:26:32 (41943): called boinc_finish EDIT: boinc pass "device 0" instead of "device 1" boinc_project 42690 0.3 0.0 2447260 3588 ?? SN 1:01AM 0:00.02 milkyway_1.37_x86_64-apple-darwin__opencl_nvidia_101 -f -np 20 -p 0.507773611546175 5.314882185745 -1.78457837505283 192.629590982596 39.017990164971 1.87224701958271 3.15095839718822 4.18477040237846 -0.533261040271202 200.315243139419 22.88 1.901 2.99 24.2836812450552 -0.4582270374547 194.265010250273 11.7345663011072 2.75052293940004 0.026636379053808 6.0638655173807 --device 0 with device 1 it's starts fine (manually) |
Send message Joined: 8 Apr 09 Posts: 70 Credit: 11,027,167,827 RAC: 0 |
Hello mancausoft, i think you are running OpenCL on your Nvidia GPU (GTX 750M Device 0) but all seem to error out. I have no experience with Darwin as an operating system, but i think it could be a driver issue, or some permissions are missing so Boinc is not allowed to use all the files it needs. Your device 1 (second device) is you Intel GPU integrated in you CPU. But this one does not have a double precision compute capability. Look for: "Device doesn't support double precision" in your last post. |
Send message Joined: 31 Aug 16 Posts: 2 Credit: 1,503,609 RAC: 0 |
For my case the problem seems in the Boinc (7.6.22) Hw detector (or opencl library). I have two card and one platform. Platform 0 Card 0 is the Intel Card Platform 0 Card 1 is the Nvidia card Mer 7 Set 02:11:47 2016 | | OpenCL: NVIDIA GPU 0: GeForce GT 750M (driver version 10.10.13 310.42.25f01, device version OpenCL 1.2, 2048MB, 2048MB available, 178 GFLOPS peak) Mer 7 Set 02:11:47 2016 | | OpenCL: Intel GPU 0: Iris Pro (driver version 1.2(Aug 29 2016 22:20:39), device version OpenCL 1.2, 1536MB, 1536MB available, 384 GFLOPS peak) And in the file coproc_info.xml I have device_num 0 for all two cards: <nvidia_opencl> <name>GeForce GT 750M</name> <vendor>NVIDIA</vendor> <device_num>0</device_num> </nvidia_opencl> <intel_gpu_opencl> <name>Iris Pro</name> <vendor>Intel</vendor> <device_num>0</device_num> </intel_gpu_opencl> all two card is detected as device 0, and boinc start: "milkyway_1.37_x86_64-apple-darwin__opencl_nvidia_101" passing the params "-device 0" instead of "-device 1" I try to start manually apple-darwin__opencl_nvidia_101 passing -device 1 and it works. Same problem with boinc 7.6.33 |
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
Hey Tim, That may be a sign that you can run more than one work unit without slowing your work units down. Have you tried running multiple work units on your GPU concurrently? Mancausoft, Our application does not run on Intel integrated GPUs as they do not properly support double precision. Everyone, Also I am still trying to get the BOINC libraries cross compiling for windows. There were a lot of changes in the BOINC libraries over the last few years and getting the kinks worked out with our build system has been a bumpy road. Jake |
Send message Joined: 13 Sep 08 Posts: 12 Credit: 131,420,119 RAC: 0 |
|
Send message Joined: 28 Sep 13 Posts: 4 Credit: 6,500,579 RAC: 0 |
|
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
apohawk, Our application does currently support your GPU, the issue is our server does not recognize it as a GPU that can crunch our work units. You can fix this short term buy using the config files others have posted on number crunching. I am hoping that the release I am working on now will fix the issue, but I can not make any promises. If it doesn't, I will keep looking for a solution. Tim, That is awesome! I am seeing your "Measured floating point speed" being reported as "3216.12 million ops/sec." Looks good to me! Jake |
Send message Joined: 13 Sep 08 Posts: 12 Credit: 131,420,119 RAC: 0 |
apohawk, Thanks. Yes, i crunched few WUs i had in queue before removing 6950 and they were crunched without any changes, app_info, any problems and validated properly. Any ETA on that release you're working on? |
Send message Joined: 22 Jan 08 Posts: 29 Credit: 242,730,423 RAC: 0 |
I've always gotten the impression that if you want to crunch for MilkyWay, it is best to run Nvidia cards... Well, if double precision is of importance, you should never choose NVIDIA cards but instead AMD as these are significantly more potent in this discipline. To the best of my knowledge, the best model on the consumer market still is the R9 280X (and here e.g. the Toxic version from Sapphire). An exception from the rule is the NVIDIA Titan Black series but that one is so expensive that you can afford many 280X cards for the price of one Titan Black such that these will again outperform the Titan Black. Also, there are no significant driver issues with AMD graphics cards as long as you are using Windows as OS. The core problem with MW is that it will just not recognize some of the AMD cards properly while other projects have absolutely no problem doing that (Einstein, Collatz, POEM, Primegrid, SETI tested). Michael. President of Rechenkraft.net e.V. - This planet's first and largest distributed computing organization. |
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
Hey apohawk, Not ETA right now. I was hoping to have it done last week, but I've run into a few issues and have been distracted a bit working on a few unrelated things. I will keep working at it and hopefully I can have this done soon. Jake |
Send message Joined: 8 Apr 09 Posts: 70 Credit: 11,027,167,827 RAC: 0 |
Thanks for the info Jake :) apohawk i can confirm the HD 6950 running well with the latest driver for it and with the current Milkyway workunits. |
Send message Joined: 22 Mar 09 Posts: 1 Credit: 199,954,756 RAC: 0 |
I am computing with three ATI 5970's, not overclocked, cooled adequately, with no overclocking on the CPU, no bugs in the RAM, or HD's. Attempting to process the WU's invariably end in computation errors, and I have selected all of the different GPU WU options one at a time, but this invariably fails. Drivers aren't an issue, I've run a clean install and the machine is stably with games and CPU-Z. This is also the case for the Collitz Conjecture project, but now Moo! Wrapper, strangely. Below, please find the lines from the Milky Way WU's. This issue persisted with the newest version of BOINC, so I reverted to an older one, 6.12 and the same result. Thank you for your time and kind attention to this matter, I hope to assist in resolving the issue! 9/11/2016 8:44:08 AM | | Starting BOINC client version 6.12.33 for windows_x86_64 9/11/2016 8:44:08 AM | | log flags: file_xfer, sched_ops, task 9/11/2016 8:44:08 AM | | Libraries: libcurl/7.19.7 OpenSSL/0.9.8l zlib/1.2.5 9/11/2016 8:44:08 AM | | Data directory: C:\ProgramData\BOINC 9/11/2016 8:44:08 AM | | Running under account HPC Multipurpose 9/11/2016 8:44:08 AM | | Processor: 12 GenuineIntel Intel(R) Core(TM) i7 CPU X 980 @ 3.33GHz [Family 6 Model 44 Stepping 2] 9/11/2016 8:44:08 AM | | Processor: 256.00 KB cache 9/11/2016 8:44:08 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 nx lm vmx tm2 popcnt aes pbe 9/11/2016 8:44:08 AM | | OS: Microsoft Windows 7: Professional x64 Edition, Service Pack 1, (06.01.7601.00) 9/11/2016 8:44:08 AM | | Memory: 23.99 GB physical, 58.17 GB virtual 9/11/2016 8:44:08 AM | | Disk: 148.95 GB total, 100.84 GB free 9/11/2016 8:44:08 AM | | Local time is UTC -5 hours 9/11/2016 8:44:08 AM | | ATI GPU 0: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1607, 1024MB, 2320 GFLOPS peak) 9/11/2016 8:44:08 AM | | ATI GPU 1: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1607, 1024MB, 2320 GFLOPS peak) 9/11/2016 8:44:08 AM | | ATI GPU 2: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1607, 1024MB, 2320 GFLOPS peak) 9/11/2016 8:44:08 AM | | ATI GPU 3: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1607, 1024MB, 2320 GFLOPS peak) 9/11/2016 8:44:08 AM | | ATI GPU 4: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1607, 1024MB, 2320 GFLOPS peak) 9/11/2016 8:44:08 AM | | ATI GPU 5: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1607, 1024MB, 2320 GFLOPS peak) 9/11/2016 8:44:08 AM | Asteroids@home | URL http://asteroidsathome.net/boinc/; Computer ID 345227; resource share 100 9/11/2016 8:44:08 AM | Collatz Conjecture | URL https://boinc.thesonntags.com/collatz/; Computer ID 260666; resource share 300 9/11/2016 8:44:08 AM | LHC@home 1.0 | URL http://lhcathomeclassic.cern.ch/sixtrack/; Computer ID 10398898; resource share 100 9/11/2016 8:44:08 AM | Milkyway@Home | URL http://milkyway.cs.rpi.edu/milkyway/; Computer ID 619562; resource share 325 9/11/2016 8:44:08 AM | Moo! Wrapper | URL http://moowrap.net/; Computer ID 593181; resource share 400 9/11/2016 8:44:08 AM | World Community Grid | URL http://www.worldcommunitygrid.org/; Computer ID 3321002; resource share 100 9/11/2016 8:44:08 AM | Collatz Conjecture | General prefs: from Collatz Conjecture (last modified 09-Sep-2016 01:31:37) 9/11/2016 8:44:08 AM | Collatz Conjecture | Host location: none 9/11/2016 8:44:08 AM | Collatz Conjecture | General prefs: using your defaults 9/11/2016 8:44:08 AM | | Reading preferences override file 9/11/2016 8:44:08 AM | | Preferences: 9/11/2016 8:44:08 AM | | max memory usage when active: 12283.56MB 9/11/2016 8:44:08 AM | | max memory usage when idle: 22110.41MB 9/11/2016 8:44:08 AM | | max disk usage: 0.00GB 9/11/2016 8:44:08 AM | | don't compute while active 9/11/2016 8:44:08 AM | | don't use GPU while active 9/11/2016 8:44:08 AM | | suspend work if non-BOINC CPU load exceeds 40 % 9/11/2016 8:44:08 AM | | (to change preferences, visit the web site of an attached project, or select Preferences in the Manager) 9/11/2016 8:44:08 AM | | Not using a proxy 9/11/2016 8:44:09 AM | | Suspending computation - computer is in use 9/11/2016 8:44:37 AM | Milkyway@Home | resumed by user 9/11/2016 8:44:46 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786046_0 using milkyway version 136 9/11/2016 8:44:46 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786045_0 using milkyway version 136 9/11/2016 8:44:46 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786042_0 using milkyway version 136 9/11/2016 8:44:46 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786041_0 using milkyway version 136 9/11/2016 8:44:46 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786039_0 using milkyway version 136 9/11/2016 8:44:46 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786031_0 using milkyway version 136 9/11/2016 8:44:47 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786046_0 finished 9/11/2016 8:44:47 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786045_0 finished 9/11/2016 8:44:47 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786042_0 finished 9/11/2016 8:44:47 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786041_0 finished 9/11/2016 8:44:47 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786039_0 finished 9/11/2016 8:44:47 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786031_0 finished 9/11/2016 8:44:47 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786030_0 using milkyway version 136 9/11/2016 8:44:47 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786029_0 using milkyway version 136 9/11/2016 8:44:47 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786025_0 using milkyway version 136 9/11/2016 8:44:47 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786024_0 using milkyway version 136 9/11/2016 8:44:47 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786023_0 using milkyway version 136 9/11/2016 8:44:47 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786022_0 using milkyway version 136 9/11/2016 8:44:48 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786030_0 finished 9/11/2016 8:44:48 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786029_0 finished 9/11/2016 8:44:48 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786025_0 finished 9/11/2016 8:44:48 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786024_0 finished 9/11/2016 8:44:48 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786023_0 finished 9/11/2016 8:44:48 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786022_0 finished 9/11/2016 8:44:48 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786021_0 using milkyway version 136 9/11/2016 8:44:48 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786020_0 using milkyway version 136 9/11/2016 8:44:48 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786019_0 using milkyway version 136 9/11/2016 8:44:48 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786017_0 using milkyway version 136 9/11/2016 8:44:48 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786016_0 using milkyway version 136 9/11/2016 8:44:48 AM | Milkyway@Home | Starting task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786014_0 using milkyway version 136 9/11/2016 8:44:49 AM | | Suspending computation - computer is in use 9/11/2016 8:44:49 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786021_0 finished 9/11/2016 8:44:49 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786020_0 finished 9/11/2016 8:44:49 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786019_0 finished 9/11/2016 8:44:49 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786017_0 finished 9/11/2016 8:44:49 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786016_0 finished 9/11/2016 8:44:49 AM | Milkyway@Home | Computation for task de_modfit_fast_15_3s_136_fixedangles3_1_1471352126_31786014_0 finished |
Send message Joined: 6 Apr 12 Posts: 42 Credit: 3,215,609 RAC: 0 |
hello my graphics cards are not happy. GIGABYTE GV-R487D5-1GD I do not use crossfire as it doesn't help. quoting my moist recent log: 13/09/2016 4:31:54 PM | | cc_config.xml not found - using defaults |
Send message Joined: 28 Jun 09 Posts: 1 Credit: 27,935,429 RAC: 0 |
I can confirm that under nvidia driver 372.70 2 1070's,3 970's and 2 960's are working fine. |
Send message Joined: 6 Apr 12 Posts: 42 Credit: 3,215,609 RAC: 0 |
The driver is the latest from Microsoft so i'm wondering is it the driver? if so I would appreciate if someone could point out a working version for me. These cards have serious grunt. |
©2024 Astroinformatics Group