| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Beta Testing
Forum: Beta Test Support Forum
Thread: Beta test For HCC - GPU v 6.50 |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 50
|
|
| Author |
|
|
ThreadRipper
Veteran Cruncher Sweden Joined: Apr 26, 2007 Post Count: 1317 Status: Offline Project Badges: 
|
I got like 75 WUs this round and they all are either valid or PV. Looking good! Boinc manager v 7.0.28. HD7970 @ Stock.
---------------------------------------- Join The International Team: https://www.worldcommunitygrid.org/team/viewTeamInfo.do?teamId=CK9RP1BKX1 AMD TR2990WX @ PBO, 64GB Quad 3200MHz 14-17-17-17-1T, RX6900XT @ Stock AMD 3800X @ PBO AMD 2700X @ 4GHz |
||
|
|
katoda
Senior Cruncher Poland Joined: Apr 28, 2007 Post Count: 170 Status: Offline Project Badges: 
|
Received around 25 of them on two machines - Vista 32bit with nVidia GTX670 and Windows7 64bit with nVidia GT330M. On the first machine, all tasks completed successfully, however I noticed stuttering while GPU part of computation.
----------------------------------------On the other machine all of the tasks (3) resulted in error - exit code 233 (0xe9). Here is the log: Result Name: BETA_ X0000038270193200409250027_ 2-- <core_client_version>6.10.58</core_client_version> <![CDATA[ <message> Na drugim ko�cu potoku nie ma �adnego procesu. (0xe9) - exit code 233 (0xe9) </message> <stderr_txt> Commandline: projects/www.worldcommunitygrid.org/wcg_beta3_img_6.50_windows_intelx86__nvidia_hcc1 X0000038270193200409250027.jp2 --device 0 <app_init_data> <major_version>6</major_version> <minor_version>10</minor_version> <release>58</release> <app_version>650</app_version> <app_name>beta3</app_name> <project_preferences> <color_scheme>Tahiti Sunset</color_scheme> <max_frames_sec>7</max_frames_sec> <max_gfx_cpu_pct>5.0</max_gfx_cpu_pct> </project_preferences> <project_dir>C:\ProgramData\BOINC/projects/www.worldcommunitygrid.org</project_dir> <boinc_dir>C:\ProgramData\BOINC</boinc_dir> <wu_name>BETA_X0000038270193200409250027</wu_name> <comm_obj_name>boinc_2</comm_obj_name> <slot>2</slot> <wu_cpu_time>0.000000</wu_cpu_time> <starting_elapsed_time>0.000000</starting_elapsed_time> <using_sandbox>0</using_sandbox> <user_total_credit>3704885.696499</user_total_credit> <user_expavg_credit>4951.823444</user_expavg_credit> <host_total_credit>163670.612947</host_total_credit> <host_expavg_credit>158.902002</host_expavg_credit> <resource_share_fraction>1.000000</resource_share_fraction> <checkpoint_period>60.000000</checkpoint_period> <fraction_done_start>0.000000</fraction_done_start> <fraction_done_end>1.000000</fraction_done_end> <gpu_type></gpu_type> <gpu_device_num>-1</gpu_device_num> <gpu_opencl_dev_index>-1</gpu_opencl_dev_index> <ncpus>0.000000</ncpus> <rsc_fpops_est>13814991649664.000000</rsc_fpops_est> <rsc_fpops_bound>276299832993280.000000</rsc_fpops_bound> <rsc_memory_bound>78643200.000000</rsc_memory_bound> <rsc_disk_bound>50000000.000000</rsc_disk_bound> <computation_deadline>1347629778.000000</computation_deadline> <vbox_window>0</vbox_window> </app_init_data> INFO: gpu_type not found in init_data.xml. INFO: GPU device not specified in init_data.xml. Checking Commandline. Boinc requested NVIDIA gpu device number 0 Found compute platform NVIDIA Corporation Selecting this platform CL_DEVICE_NAME: GeForce GT 330M CL_DEVICE_VENDOR: NVIDIA Corporation CL_DEVICE_VERSION: 260.99 CL_DEVICE_MAX_COMPUTE_UNITS: CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS: 3 CL_DEVICE_MAX_WORK_ITEM_SIZES: 512 / 512 / 64 CL_DEVICE_MAX_WORK_GROUP_SIZE: 512 CL_DEVICE_MAX_CLOCK_FREQUENCY: 1265 MHz CL_DEVICE_ADDRESS_BITS: 32 CL_DEVICE_MAX_MEM_ALLOC_SIZE: 246 MByte CL_DEVICE_GLOBAL_MEM_SIZE: 986 MByte CL_DEVICE_ERROR_CORRECTION_SUPPORT: no CL_DEVICE_LOCAL_MEM_TYPE: local CL_DEVICE_LOCAL_MEM_SIZE: 16 KByte CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE: 64 KByte CL_DEVICE_QUEUE_PROPERTIES: CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE CL_DEVICE_QUEUE_PROPERTIES: CL_QUEUE_PROFILING_ENABLE CL_DEVICE_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 cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics CL_DEVICE_COMPUTE_CAPABILITY_NV: 1.2 CL_DEVICE_REGISTERS_PER_BLOCK_NV: 16384 CL_DEVICE_WARP_SIZE_NV: 32 CL_DEVICE_GPU_OVERLAP_NV: CL_TRUE CL_DEVICE_KERNEL_EXEC_TIMEOUT_NV: CL_TRUE CL_DEVICE_INTEGRATED_MEMORY_NV: CL_FALSE Estimated kernel execution time = 2.51139 [sec] ERROR: Kernel execution time estimate too high, exiting. 05:36:44 (166680): called boinc_finish </stderr_txt> Same issue for BETA_ X0000038270191200409250027_ 3-- and BETA_ X0000038271358200409250006_ 2--  |
||
|
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1259 Status: Recently Active Project Badges:
|
What are the best settings to have a chance at getting a beta? I'm using Boinc 7.0.28 I just returned BETA_ X0000038270892200409250014-- 201209132116350_ 4 650 Valid 9/14/12 07:37:50 9/14/12 07:41:32 0.05 56.3 / 56.2 ran on my GTX 470 FYI another person returned as vaild but 3 copies errored out  |
||
|
|
jonnieb-uk
Ace Cruncher England Joined: Nov 30, 2011 Post Count: 6105 Status: Offline Project Badges: 
|
Got 8 WUs. Running 6.10.58 and AMD BeaverCreek.
----------------------------------------1 WU completed and validated. 7 WUs ran to 99.415% and hung. In each case waited at least 15min. No activity on GPU nor on the reserved CPU core. Suspended then resumed each WU in turn. Appeared to be the equivalent of a reset. Each WU ran again until 99.415% before hanging. Noticeably jerky whilst the GPU portion was running. Had the same problem with CPU tasks being downloaded whilst GPU Beta active despite CPU cach being full, as described by wplachy above. Noted the comments about Beta's running successfully on 7.0.28, but AFAIK 6.10.58 is still the latest WCG approved version. Is this advice no longer valid?? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
What are the best settings to have a chance at getting a beta? I'm using Boinc 7.0.28 The question requires qualification. In general or GPU per se? Many have a profile reserved and set to only fetch work from an intermittent project + beta participation. That practically ensures when a Beta is announced only work fetches succeed for Beta jobs, if in the feeder. If you only want GPU beta work, set the web device profile to disallow CPU work fetches. Beta tasks have a higher priority in the feeder. When a client asks for work, and a device is enrolled in the program, the server scheduler first checks if there is beta work before checking on regular production work. The client will never tell you it failed to get Beta work... it's a silent check and only informs on regular production work fetch fail or success. In this case it would then tell there is no work for DDDT2 and/or C4CW, the only intermittent projects at this time. A side question for knreed et al: There was a redundancy question on HCC by CandymanWCG. Will the HCC on GPU run in quorum 2? Vaguely remember a comment having been made on this subject by WCG tech or a project scientists. (Aging prevents me to recollect what it exactly was.) |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello Jim1348
Reference: Jim1348 Sep 14, 2012 2:03:18 AM] post I see broadly two approaches for crunchers participating in multiple grids: the one-machine-per-grid approach, and the one-machine-across-multiple-grids approach. It seems to me that the one-machine-per-grid approach stands the best chance of reigning in the scope of uncertainties arising from the interplay of GPUs and CPUs, not to mention shielding machines from having to address all possible effects of combinations of operational peculiarities/constraints of each grid at the same time -- which often ends up as a compromised solution/setting. ; |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges: 
|
andzgrid,
----------------------------------------I have been checking out the possibilities the last few weeks waiting for HCC to show up, and what works for me on this machine (may not apply to all hardware of course) is to run only GPU projects, and not mixed CPU/GPU projects. But then, I can run both POEM and HCC on my two HD 7770s without interference. Either one is running on a given card, or the other one is, but not both. In my case, I prefer HCC, but don't want to leave the cards idle waiting for the betas to show up, so I set the resource share of POEM to "1", while leaving it at "100" for WCG. That way, I get POEM jobs at the normal rate when there is no WCG work for the GPU, and thus far it has picked up the betas for HCC when they are available. We will see how it works when the full project starts. By the way, I am on BOINC 7.0.33, and will be upgrading to 7.0.36 shortly (http://boinc.berkeley.edu/dl/). They keep uncovering new bugs related to scheduling and GPUs all the time, so it is worth the risk of a new version in this case. [Edit 1 times, last edit by Jim1348 at Sep 14, 2012 10:56:55 AM] |
||
|
|
ThreadRipper
Veteran Cruncher Sweden Joined: Apr 26, 2007 Post Count: 1317 Status: Offline Project Badges:
|
I am just wondering...when we go to production with this, will it be possible to select to crunch only GPU WUs for HCC?
----------------------------------------For instance say I want to run HPF2 on the CPU cores and HCC GPU-only on my GPU. Will it be possible, so that I do not have to get CPU WUs for HCC then? Join The International Team: https://www.worldcommunitygrid.org/team/viewTeamInfo.do?teamId=CK9RP1BKX1 AMD TR2990WX @ PBO, 64GB Quad 3200MHz 14-17-17-17-1T, RX6900XT @ Stock AMD 3800X @ PBO AMD 2700X @ 4GHz |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
For instance say I want to run HPF2 on the CPU cores and HCC GPU-only on my GPU. Will it be possible, so that I do not have to get CPU WUs for HCC then? I believe they stated some time ago that if both CPU and GPU work units are available, they would send you only the GPU ones. But I don't see any specific setting for that, so whether it works that way remains to be seen. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
There is the 2 options to narrow the CPU/GPU and/or selection:
1) If my computer can process work on my graphics card, then please send me work to run on my graphics card for the projects that I have selected above. 2) "Allow research to run on my CPU?" Yes/No Latter is broken at this time it appears from various reports [changes don't stick] Thus, if you select HPF2 and GPU [not HCC] and you'd combine that with the option: 3) If there is no work available for my computer for the projects I have selected above, please send me work from another project. Logic would be that any GPU work would be send for an idle GPU recourse, speak for HCC GPU. No testing possible as there is no GPU work in production. Rather complex, but we know that, the techs too, until they get around redesigning and make the structure more matrix oriented. |
||
|
|
|