Posts by Morten Ross
|
1)
Message boards :
Number crunching :
Computation errors.
(Message 60141)
Posted 11 Oct 2013 by Morten Ross Post: One of my hosts started experiencing computation errors one day ago. The first change is that all tasks stay 10 or more seconds at 100% complete. M@H Separation completes successfully, but computation times have increased due to the added 10+ seconds at 100% complete. Thus I'm only able to crunch M@H Separation. I've detached and reattached to project - no change. Only system change was Windows Update security patches on Wednesday. Error excerpt: Failed to move file 'separation_checkpoint_tmp' to 'separation_checkpoint' (6801): Transaction support within the specified resource manager is not started or was shut down due to an error. Failed to update checkpoint file ('separation_checkpoint_tmp' to 'separation_checkpoint') (2): No such file or directory Failed to write final checkpoint Failed to calculate likelihood |
|
2)
Message boards :
Number crunching :
ATI tasks running on Nvidia!
(Message 55596)
Posted 21 Sep 2012 by Morten Ross Post: Re-ran the OpenCl.msi and problem was resolved. |
|
3)
Message boards :
Number crunching :
ATI tasks running on Nvidia!
(Message 55594)
Posted 21 Sep 2012 by Morten Ross Post: The host has one GTX 690 (Seti only) and one HD 5970 (Milkyway only) and see that Milkyway (or BOINC?) is not assigning the ATI tasks to the HD 5970, and I have no idea how to correct this. My project settings has ATI-only, despite this, it is assigning OpenCL tasks to the GTX 690. BOINC reports this: 21/09/2012 19:11:56 NVIDIA GPU 0: GeForce GTX 690 (driver version 30623, CUDA version 5000, compute capability 3.0, 2048MB, 522 GFLOPS peak) 21/09/2012 19:11:56 NVIDIA GPU 1: GeForce GTX 690 (driver version 30623, CUDA version 5000, compute capability 3.0, 2048MB, 522 GFLOPS peak) 21/09/2012 19:11:56 ATI GPU 0: ATI Radeon HD5800 series (Cypress) (CAL version 1.4.1741, 1024MB, 2320 GFLOPS peak) 21/09/2012 19:11:56 ATI GPU 1: ATI Radeon HD5800 series (Cypress) (CAL version 1.4.1741, 1024MB, 2320 GFLOPS peak) Is the only way to override default logistics to use app_info also for Milkyway? Looking at the stderr it appears that the Milkyway science app will select any available device within the platform - in this case bot ATI and Nvidia is in this platform, and for some reason Nvidia takes presedence. It's Guessing ATI is preferred: "Guessing preferred OpenCL vendor 'Advanced Micro Devices, Inc.'", but it's omitted..... Stderr: <search_application> milkyway_separation 1.02 Windows x86_64 double OpenCL </search_application> Unrecognized XML in project preferences: max_gfx_cpu_pct Skipping: 100 Skipping: /max_gfx_cpu_pct Unrecognized XML in project preferences: allow_non_preferred_apps Skipping: 1 Skipping: /allow_non_preferred_apps Unrecognized XML in project preferences: nbody_graphics_poll_period Skipping: 30 Skipping: /nbody_graphics_poll_period Unrecognized XML in project preferences: nbody_graphics_float_speed Skipping: 5 Skipping: /nbody_graphics_float_speed Unrecognized XML in project preferences: nbody_graphics_textured_point_size Skipping: 250 Skipping: /nbody_graphics_textured_point_size Unrecognized XML in project preferences: nbody_graphics_point_point_size Skipping: 40 Skipping: /nbody_graphics_point_point_size Guessing preferred OpenCL vendor 'Advanced Micro Devices, Inc.' Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4' Error reading astronomy parameters from file 'astronomy_parameters.txt' Trying old parameters file Using SSE4.1 path Found 1 platform Platform 0 information: Name: NVIDIA CUDA Version: OpenCL 1.1 CUDA 4.2.1 Vendor: NVIDIA Corporation Extensions: cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_d3d9_sharing cl_nv_d3d10_sharing cl_khr_d3d10_sharing cl_nv_d3d11_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll Profile: FULL_PROFILE Didn't find preferred platform Using device 0 on platform 0 Found 2 CL devices Device 'GeForce GTX 690' (NVIDIA Corporation:0x10de) (CL_DEVICE_TYPE_GPU) Driver version: 306.23 Version: OpenCL 1.1 CUDA Compute capability: 3.0 Max compute units: 8 Clock frequency: 1019 Mhz Global mem size: 2147483648 Local mem size: 49152 Max const buf size: 65536 Double extension: cl_khr_fp64 Build log: -------------------------------------------------------------------------------- ptxas : info : Compiling entry function 'probabilities' for 'sm_30' ptxas : info : Function properties for probabilities 88 bytes stack frame, 84 bytes spill stores, 84 bytes spill loads ptxas : info : Used 62 registers, 388 bytes cmem[0], 176 bytes cmem[2] -------------------------------------------------------------------------------- Build log: -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Estimated Nvidia GPU GFLOP/s: 522 SP GFLOP/s, 65 DP FLOP/s Using a target frequency of 60.0 Using a block size of 8192 with 5 blocks/chunk Using clWaitForEvents() for polling with initial wait of 12 ms (mode 0) Range: { nu_steps = 640, mu_steps = 1600, r_steps = 1400 } Iteration area: 2240000 Chunk estimate: 53 Num chunks: 55 Chunk size: 40960 Added area: 12800 Effective area: 2252800 Initial wait: 12 ms |
|
4)
Message boards :
Number crunching :
Still not getting ATI tasks
(Message 42648)
Posted 6 Oct 2010 by Morten Ross Post: It was caused by BOINC version 6.11.9. I reverted back to 6.10.56 and tasks are again downloaded. |
|
5)
Message boards :
Number crunching :
Still not getting ATI tasks
(Message 42639)
Posted 6 Oct 2010 by Morten Ross Post: As you can see from host that has allready been done.... |
|
6)
Message boards :
Number crunching :
Still not getting ATI tasks
(Message 42626)
Posted 6 Oct 2010 by Morten Ross Post: I can see from other hosts that new ATI tasks are distributed. I still get nothing - "Scheduler request completed: got 0 tasks". I had a disk crash last week and the host lost 48 tasks - the only reason I can come up with why I'm not getting any, but I'm not aware there is a mechanism to send zero tasks to such hosts. |
|
7)
Message boards :
Number crunching :
Why is 5970 classified as 5800 series in BOINC?
(Message 37709)
Posted 24 Mar 2010 by Morten Ross Post: Boinc has for a long time correctly classified Nvidias old 5970 counterpart GTX 295 as just that. GTX 295 is in fact 2x GTX 280, and as I recall has never been classified as GTX 280 by Boinc. Perhaps the Boinc developers have a preference for Nvidia ;-) ? |
|
8)
Message boards :
Number crunching :
Receving 24 WUs - should get 48
(Message 37708)
Posted 24 Mar 2010 by Morten Ross Post: That is much better than only getting 6 each due to only having one GPU. Running MW on CPU is really a waste of electricity :-) I'm used to Seti@Home, and they adhere to the Your account "Computing Settings" for BOINC. So if I have set "5 days" for the "Maintain enough work for an additional", Boinc will download WUs to maintain a cache for about 9 days. This ensures WUs also for most of the maintenance and downtimes that occur. MW has instead configured all clients to query their servers extremely frequently instead, (some clients will query and download once a minute) and for a very low cache of WUs. |
|
9)
Message boards :
Number crunching :
Receving 24 WUs - should get 48
(Message 37657)
Posted 22 Mar 2010 by Morten Ross Post: D'oh! :-) As I am in "GPU-land" and have also selected, and only receive GPU work, I hadn't considered that I still would be calculated by CPU :-( Veeeery logical...not.... Thanks for your clarification :-) Morten |
|
10)
Message boards :
Number crunching :
Why is 5970 classified as 5800 series in BOINC?
(Message 37654)
Posted 22 Mar 2010 by Morten Ross Post: Purely asthetic - they have the same GPU, just that the 5970 has two of them. Its early days for software writers to catch up purely on labels. Give it another couple of months and no doubt they will. At the moment BOINC shows a 5970 as [2]5870's in the computer summary page. If this is so, then someone is hardcoding the info. Best practice is reading from the card or operating system. This will get the correct info. Even a pre-5970 GPU-Z version will correctly read 5900 series.... |
|
11)
Message boards :
Number crunching :
Receving 24 WUs - should get 48
(Message 37651)
Posted 22 Mar 2010 by Morten Ross Post: Hi, I have 2 computers each running one 5970. One host receives 48 WUs, the other only 24. Both cards are dedicated to MW. Why doesn't both receive 48 WUs? Morten |
|
12)
Message boards :
Number crunching :
Why is 5970 classified as 5800 series in BOINC?
(Message 37650)
Posted 22 Mar 2010 by Morten Ross Post: Hi, Currently BOINC classifies 5970 as: 22/03/2010 02:16:06 ATI GPU 0: ATI Radeon HD5800 series (Cypress) (CAL version 1.4.556, 1024MB, 2720 GFLOPS peak) It should read; 22/03/2010 02:16:06 ATI GPU 0: ATI Radeon HD5900 series (Cypress) (CAL version 1.4.556, 1024MB, 2720 GFLOPS peak) I haven't found any utility that does this kind of classification (GPU-Z, MSI Afterburner or ATI CCC). What is the basis for this, and how to correct? Morten |
|
13)
Message boards :
Number crunching :
5970 - only one GPU available in BOINC
(Message 37554)
Posted 19 Mar 2010 by Morten Ross Post: I think my problem is that this mobo is Nvidia 790i :-( As for the slots use, that's the way I've done it as well. After I got the first 5970 working, I installed the second. I got all 4 GPUs available to Boinc once with one physical monitor attached, and when I disconnected the monitor I was back to seeing only 2 GPUs no matter what. I'm back on trying dummy plug again.... Do you use the xfire bridge, or software-only? Morten |
|
14)
Message boards :
Number crunching :
5970 - only one GPU available in BOINC
(Message 37534)
Posted 19 Mar 2010 by Morten Ross Post: I had to remove the Nvidia card in order for both GPUs to be usable/enable CrossfireX. I don't know why or how the Nvidia interfered with ATI, as both should work in same system... |
|
15)
Message boards :
Number crunching :
5970 - only one GPU available in BOINC
(Message 37506)
Posted 18 Mar 2010 by Morten Ross Post: Hi, I seem to have the same problem as with Nvidia GTX295/Cuda, I thought was not an issue with 5970. Latest CAL installs, both GPUs are available in Windows, but only one GPU is avaibale in BOINC. BOINC log says: "18/03/2010 18:19:38 ATI GPU 0: ATI Radeon HD5800 series (Cypress) (CAL version 1.4.553, 1024MB, 2320 GFLOPS peak) " It should also see the other GPU. Dummy plug or no dummy plug - same result. Extending or not extending desktop - same result. This is sooo frustrating, as I invested in two 5970s to move away from the hazzle of gtx295 which can only run to perfection on WinXP (Nvidia 480 needs first to be released, then drivers matured). Looking at the top hosts here, all are Win7 and 5970 (still recognized as 5800 series by BOINC), so I hope you could shed some light on this. Thanks! Morten |
©2024 Astroinformatics Group