BRP4 1.31/1.32 GPU app release: feedback thread
13 Nov 2012 13:07:02 UTC
Topic 196625
(moderation:
Please use this thread to discuss the newly released BRP4 GPU applications v1.31/v1.32.
New features and improvements:
* native Linux x86_64 apps for NVIDIA and AMD/ATI GPUs
* improved overall system stability when running our GPU tasks
Note: v1.32 is technically identical to v1.31. See below for details.
Cheers,
Oliver
Einstein@Home Project
BRP4 1.31/1.32 GPU app release: feedback thread
)
Trying with the Windows/cuda32 app:
(from http://einstein.phys.uwm.edu/host_sched_logs/5744/5744895)
That host (5744895) has successfully run BRP4 v1.28 with the current driver - I think we need an additional BRP4cuda32nv301 plan class for the new app, for drivers above the 'sleepy monitor' window.
Well, looks like there is a
)
Well, looks like there is a problem.
My nVidia system is no longer getting work for GTX550.
Message says:
2012-11-13 15:06:33.6007 [PID=26262] Request: [USER#xxxxx] [HOST#5394959] [IP xxx.xxx.xxx.162] client 7.0.31
2012-11-13 15:06:33.6081 [PID=26262] [send] effective_ncpus 4 max_jobs_on_host_cpu 999999 max_jobs_on_host 999999
2012-11-13 15:06:33.6081 [PID=26262] [send] effective_ngpus 1 max_jobs_on_host_gpu 999999
2012-11-13 15:06:33.6081 [PID=26262] [send] Not using matchmaker scheduling; Not using EDF sim
2012-11-13 15:06:33.6081 [PID=26262] [send] CPU: req 0.00 sec, 0.00 instances; est delay 0.00
2012-11-13 15:06:33.6081 [PID=26262] [send] CUDA: req 17460.00 sec, 1.00 instances; est delay 0.00
2012-11-13 15:06:33.6081 [PID=26262] [send] work_req_seconds: 0.00 secs
2012-11-13 15:06:33.6081 [PID=26262] [send] available disk 29.25 GB, work_buf_min 0
2012-11-13 15:06:33.6081 [PID=26262] [send] active_frac 0.999801 on_frac 0.370087 DCF 1.776152
2012-11-13 15:06:33.6092 [PID=26262] [send] [HOST#5394959] is reliable
2012-11-13 15:06:33.6093 [PID=26262] [send] set_trust: random choice for error rate 0.000010: yes
2012-11-13 15:06:33.6213 [PID=26262] [version] Checking plan class 'opencl-ati'
2012-11-13 15:06:33.6218 [PID=26262] [version] reading plan classes from file '../plan_class_spec.xml'
2012-11-13 15:06:33.6218 [PID=26262] [version] parsed project prefs setting 'gpu_util_brp': 0.500000
2012-11-13 15:06:33.6218 [PID=26262] [version] No ATI devices found
2012-11-13 15:06:33.6218 [PID=26262] [version] Checking plan class 'BRP4SSE'
2012-11-13 15:06:33.6218 [PID=26262] [version] project prefs setting 'also_run_cpu' (1.000000) prevents using plan class.
2012-11-13 15:06:33.6218 [PID=26262] [version] Checking plan class 'BRP4cuda32'
2012-11-13 15:06:33.6219 [PID=26262] [version] parsed project prefs setting 'gpu_util_brp': 0.500000
2012-11-13 15:06:33.6219 [PID=26262] [version] driver version required max: -29053, supplied: 30697
2012-11-13 15:06:33.6219 [PID=26262] [version] Checking plan class 'opencl-ati'
2012-11-13 15:06:33.6219 [PID=26262] [version] parsed project prefs setting 'gpu_util_brp': 0.500000
2012-11-13 15:06:33.6219 [PID=26262] [version] No ATI devices found
2012-11-13 15:06:33.6219 [PID=26262] [version] no app version available: APP#19 (einsteinbinary_BRP4) PLATFORM#9 (windows_x86_64) min_version 0
2012-11-13 15:06:33.6219 [PID=26262] [version] no app version available: APP#19 (einsteinbinary_BRP4) PLATFORM#2 (windows_intelx86) min_version 0
2012-11-13 15:06:33.6255 [PID=26262] [send] [HOST#5394959] is looking for work from a non-preferred application
2012-11-13 15:06:33.7292 [PID=26262] [version] Checking plan class 'SSE2'
2012-11-13 15:06:33.7293 [PID=26262] [version] plan class ok
2012-11-13 15:06:33.7293 [PID=26262] [version] Don't need CPU jobs, skipping version 113 for einstein_S6LV1 (SSE2)
2012-11-13 15:06:33.7293 [PID=26262] [version] no app version available: APP#20 (einstein_S6LV1) PLATFORM#9 (windows_x86_64) min_version 0
2012-11-13 15:06:33.7293 [PID=26262] [version] no app version available: APP#20 (einstein_S6LV1) PLATFORM#2 (windows_intelx86) min_version 0
2012-11-13 15:06:33.7309 [PID=26262] [debug] [HOST#5394959] MSG(high) No work sent
2012-11-13 15:06:33.7309 [PID=26262] [debug] [HOST#5394959] MSG(high) see scheduler log messages on http://einstein.phys.uwm.edu//host_sched_logs/5394/5394959
2012-11-13 15:06:33.7309 [PID=26262] Sending reply to [HOST#5394959]: 0 results, delay req 60.00
2012-11-13 15:06:33.7312 [PID=26262] Scheduler ran 0.137 seconds
I don't know which part is of interest, but I hope this can help.
Regards
Alexander
Same here, none of my systems
)
Same here, none of my systems are getting brp4cuda work units.
good morning , i dont
)
good morning ,
i dont receive any more work units for my nvidia gtx 560 Ti
Scheduler log message :
2012-11-13 14:51:18.1481 [PID=3223] Request: [USER#xxxxx] [HOST#5740399] [IP xxx.xxx.xxx.215] client 7.0.28
2012-11-13 14:51:18.1559 [PID=3223 ] [send] effective_ncpus 4 max_jobs_on_host_cpu 999999 max_jobs_on_host 999999
2012-11-13 14:51:18.1559 [PID=3223 ] [send] effective_ngpus 1 max_jobs_on_host_gpu 999999
2012-11-13 14:51:18.1559 [PID=3223 ] [send] Not using matchmaker scheduling; Not using EDF sim
2012-11-13 14:51:18.1559 [PID=3223 ] [send] CPU: req 0.00 sec, 0.00 instances; est delay 0.00
2012-11-13 14:51:18.1559 [PID=3223 ] [send] CUDA: req 21780.00 sec, 1.00 instances; est delay 0.00
2012-11-13 14:51:18.1559 [PID=3223 ] [send] work_req_seconds: 0.00 secs
2012-11-13 14:51:18.1559 [PID=3223 ] [send] available disk 99.70 GB, work_buf_min 0
2012-11-13 14:51:18.1559 [PID=3223 ] [send] active_frac 0.877768 on_frac 0.438646 DCF 1.028027
2012-11-13 14:51:18.1579 [PID=3223 ] [send] [HOST#5740399] is reliable
2012-11-13 14:51:18.1580 [PID=3223 ] [send] set_trust: random choice for error rate 0.000010: yes
2012-11-13 14:51:18.2560 [PID=3223 ] [version] Checking plan class 'SSE2'
2012-11-13 14:51:18.2564 [PID=3223 ] [version] reading plan classes from file '../plan_class_spec.xml'
2012-11-13 14:51:18.2564 [PID=3223 ] [version] plan class ok
2012-11-13 14:51:18.2564 [PID=3223 ] [version] Don't need CPU jobs, skipping version 113 for einstein_S6LV1 (SSE2)
2012-11-13 14:51:18.2564 [PID=3223 ] [version] no app version available: APP#20 (einstein_S6LV1) PLATFORM#2 (windows_intelx86) min_version 0
2012-11-13 14:51:18.2688 [PID=3223 ] [version] Checking plan class 'BRP4SSE'
2012-11-13 14:51:18.2689 [PID=3223 ] [version] plan class ok
2012-11-13 14:51:18.2689 [PID=3223 ] [version] Don't need CPU jobs, skipping version 122 for einsteinbinary_BRP4 (BRP4SSE)
2012-11-13 14:51:18.2689 [PID=3223 ] [version] Checking plan class 'BRP4cuda32'
2012-11-13 14:51:18.2689 [PID=3223 ] [version] parsed project prefs setting 'gpu_util_brp': 1.000000
2012-11-13 14:51:18.2689 [PID=3223 ] [version] driver version required max: -29053, supplied: 30142
2012-11-13 14:51:18.2689 [PID=3223 ] [version] Checking plan class 'opencl-ati'
2012-11-13 14:51:18.2689 [PID=3223 ] [version] parsed project prefs setting 'gpu_util_brp': 1.000000
2012-11-13 14:51:18.2689 [PID=3223 ] [version] No ATI devices found
2012-11-13 14:51:18.2689 [PID=3223 ] [version] no app version available: APP#19 (einsteinbinary_BRP4) PLATFORM#2 (windows_intelx86) min_version 0
2012-11-13 14:51:18.2724 [PID=3223 ] [send] [HOST#5740399] is looking for work from a non-preferred application
2012-11-13 14:51:18.2776 [PID=3223 ] [debug] [HOST#5740399] MSG(high) No work sent
And in the Log Journal of BOINC :
2012-11-13 09:51:13 | Einstein@Home | Requesting new tasks for NVIDIA
2012-11-13 09:51:16 | Einstein@Home | Scheduler request completed: got 0 new tasks
2012-11-13 09:51:16 | Einstein@Home | No work sent
2012-11-13 09:51:16 | Einstein@Home | see scheduler log messages on http://einstein.phys.uwm.edu//host_sched_logs/5740/5740399
jeanguy
ATI systems still get
)
ATI systems still get work.
edit:
1.31 opencl-ati
RE: Trying with the
)
Right! Sorry for the inconveniences, I just de-deprecated the old 1.28 Windows app for newer drivers, we will install a new proper app version with the new executable tomorrow.
Cheers
HB
getting 1.28 wu's
)
getting 1.28 wu's now.
THX.
Alexander
same, getting units now
)
same, getting units now :)
gj team
edit;
units are taking about 10-12% longer to complete.
Yeah things are getting worse
)
Yeah things are getting worse for me. Now completing CUDA tasks in 31 mins whereas before it was 24 minutes.
Disregard, reboot fixed it.
)
Disregard, reboot fixed it. Business as usual :)