Message boards :
Number crunching :
Tasks ending in errors
Message board moderation
Previous · 1 · 2 · 3 · Next
Author | Message |
---|---|
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
Thanks Gant(SSSF), as you said: With the CPU core reserved, it should run 3 (or even 4 WUs) at a time without things being laggy (even 0.4 or 0.3 may be enough). ~~~~~~~~~~~~ it runs 4 WUs great. |
Al Send message Joined: 3 Apr 99 Posts: 1682 Credit: 477,343,364 RAC: 482 |
Hey heinz, quick note to say congratz for pulling the old boy back out of storage and getting it fired back up & into useful, productive work! |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
BOINC Version 7.6.22 (x64) V8-Xeon run 0.2 per GPU device 0 runs 5 wu's device 1 runs 5 wu's device 2 runs still 4 wu' together 14 wu's are running, question is why run device 2 still 4 and not 5 wu's as you can here see --> v8-xeon_runs_14_wus have a look here --> v8-xeon_0.20_gpu what is missing ? here is app Version <app> <name>setiathome_v8</name> </app> <file_info> <name>Lunatics_x41zi_win32_cuda50.exe</name> <executable/> </file_info> <file_info> <name>cudart32_50_35.dll</name> <executable/> </file_info> <file_info> <name>cufft32_50_35.dll</name> <executable/> </file_info> <file_info> <name>mbcuda.cfg</name> </file_info> <app_version> <app_name>setiathome_v8</app_name> <version_num>800</version_num> <platform>windows_intelx86</platform> <api_version>6.2.18</api_version> <plan_class>cuda50</plan_class> <avg_ncpus>0.040000</avg_ncpus> <max_ncpus>1.000000</max_ncpus> <coproc> <type>CUDA</type> <count>0.2</count> </coproc> <file_ref> <file_name>Lunatics_x41zi_win32_cuda50.exe</file_name> <main_program/> </file_ref> <file_ref> <file_name>cudart32_50_35.dll</file_name> </file_ref> <file_ref> <file_name>cufft32_50_35.dll</file_name> </file_ref> <file_ref> <file_name>mbcuda.cfg</file_name> </file_ref> </app_version> D5400XS V8-Xeon |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Hi Heinz, Could be boinc client memory or disk limits ? "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
Hi Heinz, I give him 16 cpus, but it does not help ~~~~~~~~~~~~~~~~~~~~~~~~ 15.08.2016 00:25:27 | | Processor: 16 GenuineIntel Intel(R) Xeon(R) CPU E5405 @ 2.00GHz [Family 6 Model 23 Stepping 6] 15.08.2016 00:25:27 | | 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 syscall nx lm vmx tm2 dca pbe 15.08.2016 00:25:27 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.10240.00) 15.08.2016 00:25:27 | | Memory: 16.00 GB physical, 18.37 GB virtual 15.08.2016 00:25:27 | | Disk: 931.50 GB total, 808.91 GB free Maybe BOINC limitations ? perhaps I should use any older BOINC Version |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
Surprize, with BOINC Version 7.4.42 (X64) it runs 5 wu's per GPU smoothly on the Titans. Looks like a error in the new Boinc Version v8-xeon_runs_15_wus D5400XS V8-Xeon |
rob smith Send message Joined: 7 Mar 03 Posts: 22216 Credit: 416,307,556 RAC: 380 |
...while the Titans may be running 5 concurrent tasks have you had a look at the run times? The few I looked at are long, in the range 90mins to >150mins. (With three concurrent on a GTX980 I was seeing times between 40 and 50 minutes) Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
ok, looked up I use v8-xeon for work too. Therefore I have separate settings for work as shown: If the Computer will be used, use max 50% If the Computer will be not used, use max 90% SWAP: use max 75% ~~~~~~~~~~~~~~~~~~~ This way I can work with the machine while crunching. I get a lot of the vlar Guppis too. Sure with 5 per GPU runtime increase. My goal is to test how it is possible smoothly run 4, 5 or more wu's per GPU on v8-xeon. Results on --> hostid=6944847 D5400XS V8-Xeon |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
Let the Cache now drain out to test other Settings. Sicherung der Aufgaben höchstens alle 300 Sekunden ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ this prevent the high Bus traffic running 15 Wu's on GPU's parallel. Would I run v8-Xeon under full load, I can run 16 CPU WU's additional. This I have also tried two years ago as I got my 2 Mrd credits. D5400XS V8-Xeon |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
I started new and set 32 CPU's for full load, It runs now with 32 CPU-wu's paralllel so far ok. This streess My FBDIMMs high, Temps on FBDIMMS are ~90 grd Celsius. Normally V8-Xeon run parallel Work on it's GPU's, but till now no GPU start a gpu-wu. looking now at all the settings again edit: looked up and found nothing, the difference to my last Settings for work is I have allowed CPU work --> Ressourcenaufteilung 100 CPU benutzen yes ATI-Grafikprozessor benutzen no NVIDIA-Grafikprozessor benutzen yes |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
Meanwhile I started new with 8 cpu's to decrease temp of the FB-Dimms. I will not cancel the cpu-work... So we must wait till all 32 cpu-wu's are ready. I set Count to 0.5 <count>0.5</count> but no success, BOINC did not start the GPU-wu's If all cpu-work is done, I will exclude CPU usage in Settings as it was before. Then we will see if it starts again... |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
No Chance, we must wait till all 32 CPU wu's are done. Tomorrow it will be ready. |
The_Matrix Send message Joined: 17 Nov 03 Posts: 414 Credit: 5,827,850 RAC: 0 |
but no success, BOINC did not start the GPU-wu's Benutzt du denn KEINE app_config.xml Datei in dem Project-Ordner von SETI ? <app_config> <app> <name>setiathome_v8</name> <max_concurrent>20</max_concurrent> <gpu_versions> <gpu_usage>1.0</gpu_usage> <cpu_usage>.25</cpu_usage> </gpu_versions> </app> </app_config> So schaut meine aus. Für max_concurrent würde ich dir 50 empfehlen, das ist die maximale Anzahl der Workunits die gleichzeitig bearbeitet werden können.Grüße p.s. gpu/cpu_usage nach eigenem Ermessen |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
Ok, app_config.xml hatte ich noch nicht. hatte immer alles in cc_config.xml. hat aber gefunzt. Auf jeder GPU 1 WU. sieht so aus als wenn <gpu_usage>1.0</gpu_usage> das macht, obwohl in cc_config <count>0.2</count> angegeben ist. MfG _heinz |
The_Matrix Send message Joined: 17 Nov 03 Posts: 414 Credit: 5,827,850 RAC: 0 |
ja das ist so mit <gpu_usage> Lade dir die Freeware GPU-Z und schau dir die GPU Auslastung mal in % an, ist die nur bei 50% , kannst gpu usage auf 0.5 setzen, bei 33% auf 0.33 setzen usw. Andere empfehlen auch ein paar CPU Kerne frei zu lassen, dazu die Prozessornutzung auf 75% schalten, da meint man man hat ja weniger Leistung und verschwendet Power, stimmt aber nicht. Das OpenCL verbraucht irgendwie neben der Prozessorlast noch irgendwelche andere Resourcen, weis jetzt nicht welche, aber dafür brauchts die freigelegte Prozessorleistung. Resultat ist eine gleichmäßigere Auslastung der GPU, allerdings verbaucht die GPU dann auch MEHR Strom. |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
With app_config.xml I get it running. 3 wu / GPU and 32 cpu-wu's run parallel. if cpu-wu's are done, I let run the machine on cpu empty. The pressure on the FB-Dimms are enormous with temps up to 98 grd Celsius. The night is cool, so it can run under full load V8-Xeon runs now 41 wu's parallel. D5400XS V8-Xeon |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
All cpu-wu's are out now, so we can came back to my normal modus, still chrunching with GPU's on V8-Xeon. It was a good experiment to run V8-Xeon under full load again. But this is more reasonable in winter times, if it is cool enough too hold the temperatures into normal limits. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Running now moderate 3 wu's / GPU so I can do my additional work without any issue with the machine too. Thank's to all for your support. :-) D5400XS V8-Xeon |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
During the weekly outage I run Primegrid PPS Sieve. Now if want work for seti I get this message: 17.08.2016 14:48:12 | SETI@home | Not requesting tasks: don't need (CPU: ; NVIDIA GPU: not highest priority project) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ How can I handle this without stopping Primegrid. I want to run seti and Primegrid together on the same machine. What I had todo Setting the priority thingy right. ? D5400XS V8-Xeon |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
short Review about the last week with Pictures: look FBDIMM2 has 98grd/Celsius v8-xeon_32cpu_wu_parallel ~~~~~~~~~~~~~~~~~~~~~~~~~ since 9 days v8-xeon run PrimeGrid on GPU v8-xeon_tophost_pg_nr_9 v8-xeon_topuser_pg_nr_36 v8-xeon_tpp_by_work24h_pg_nr_7 ~~~~~~~~~~~~~~~~~~~~~~~~~ v8-xeon runs now 9days 24h absolut stable with high continuous output into tophost list. And don't forget we have summer with hot days up to 35 grd/C here. D5400XS V8-Xeon |
The_Matrix Send message Joined: 17 Nov 03 Posts: 414 Credit: 5,827,850 RAC: 0 |
..... <gpu_versions> <max_concurrent>20</max_concurrent> .... </gpu_versions> is the right place for max_conrurrent, i were false. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.