| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project Forum
Thread: Running two CEP's does not allow 100% CPU usage |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 10
|
|
| Author |
|
|
tekennelly
Cruncher Joined: Oct 10, 2005 Post Count: 45 Status: Offline Project Badges: 
|
I have a quad core Intel running four WCG processes all the time. Whenever I get two CEP processes running the System Idle time become 4 to 5 % of one processor. The cause is probably the high page fault rate with the CEP processes. Note that the page fault rate for FAAH and the others is zero.
When I suspend one of the CEP processes and a FAAH takes over then the machine runs 100%. How can I tell BONIC to run 5 tasks so that I can keep the machine busy when two or more CEP's are running. |
||
|
|
Sid2
Senior Cruncher USA Joined: Jun 12, 2007 Post Count: 259 Status: Offline Project Badges: 
|
Which OS?
----------------------------------------. . . how much ram? My C2Q 6600, 32-bit XP, 4Gig has no issues with running 4 CEP's.  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Five tasks will probably just make the problem worse.
|
||
|
|
tekennelly
Cruncher Joined: Oct 10, 2005 Post Count: 45 Status: Offline Project Badges:
|
The OS is XP and I have 4 Gig of RAM.
Note that since the other WCG tasks are not incurring a page fault rate then I think starting another task will not hurt anything with regard to memory. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I have a Q9550 with 4GB of RAM and Vista XP Home Premium. I was running 4 CEP jobs earlier this afternoon at the same time with 100% processor usage and no problems. I didn't take a look at the page faults though. I recently changed the profile to work CEP and HCC only to build up my stats on those, and it seems to be working nicely with that. The HCC jobs come and go in about 3-4 hours, while the CEP jobs go for half a day or more sometimes.
Maybe a "waiting for memory" issue for yours? I don't think these CEP jobs are too memory intensive though... You might want to keep the CEP jobs off of that specific machine until the techs tweak the sizes of the workunits if it does not like to crunch multiple CEP jobs at the same time. Or what you could do is suspend all but one CEP job until that one finishes and start another, but that would take a lot of babysitting the machine. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Maybe one of the CEP jobs specifically is causing an unusually high amount of PFs and this is the problem?
You could try letting these finish and a new set of CEP jobs come in and see if the problem duplicates? |
||
|
|
tekennelly
Cruncher Joined: Oct 10, 2005 Post Count: 45 Status: Offline Project Badges:
|
I am probably splitting hairs so lets just drop this discussion because it is probably not worth anyone's time to think about it.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If you want to experiment, use the <ncpus> setting in cc_config.xml
|
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges: 
|
Hi !
----------------------------------------just for information ! After some days crunching for CEP, I have to notice that the efficiency variability is particular high ! (assuming that granted credits / core / hour is a reliable efficiency indicator) On a double Quad Xeon (5345) running XP 64: - NR4W provides around 130 credits / 10 hours (per core) - CEP provides between 65 and 125 credits / 10 hours (per core) Additionally to very big WUs (around 20 to 27 hours on the above mentioned host, around 50 hours on older hosts), this very big efficiency variability should be a good reason for some concerns and investigations. Have a nice day, Yves |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Very few words need spending on "value" for effort Vista-Quad v XP-C2D, former having never ran more than 2 concurrently.
----------------------------------------Quad: E000011_ 057A_ 00020u00h_ 1-- 628290 Valid 8-12-08 23:24:10 10-12-08 07:58:09 11.24 168.5 / 147.5 < E000009_ 679A_ 00012f00p_ 1-- 628290 Pending Validation 8-12-08 12:31:04 9-12-08 21:08:40 20.64 315.4 / 0.0 E000009_ 202A_ 00012400a_ 0-- 628290 Valid 8-12-08 09:00:36 9-12-08 08:41:53 15.59 238.2 / 140.2 < Outlier...no joke E000005_ 430A_ 00010a00a_ 1-- 628290 Pending Validation 7-12-08 12:17:23 8-12-08 09:08:57 12.87 196.7 / 0.0 E000003_ 198A_ 000024006_ 1-- 628290 Valid 6-12-08 19:18:30 7-12-08 19:30:16 14.56 225.1 / 178.7 < E000001_ 891A_ 000019001_ 0-- 628290 Valid 6-12-08 10:08:39 7-12-08 13:46:14 17.88 276.4 / 243.0 C2D: E000006_ 416A_ 00010x00u_ 0-- 95711 Valid 7-12-08 19:03:09 10-12-08 09:03:18 27.51 280.7 / 276.4 E000004_ 774A_ 00003500s_ 0-- 95711 Valid 7-12-08 07:26:14 9-12-08 21:41:54 29.69 302.9 / 298.4 E000003_ 392A_ 00002800w_ 0-- 95711 Valid 6-12-08 21:00:53 8-12-08 22:17:44 23.87 243.5 / 248.6 E000002_ 621A_ 00001q00h_ 0-- 95711 Valid 6-12-08 14:55:06 8-12-08 13:29:19 27.58 281.4 / 286.8 E000000_ 885A_ 00000l003_ 1-- 95711 Valid 6-12-08 02:06:26 7-12-08 21:22:41 22.55 230.1 / 258.6 One Quad Quorum: E000003_ 198A_ 000024006_ 0-- Valid 6-12-08 19:30:32 10-12-08 06:00:16 15.45 145.7 / 178.7 < VVat? E000003_ 198A_ 000024006_ 1-- Valid 6-12-08 19:18:30 7-12-08 19:30:16 14.56 225.1 / 178.7 One Quad Outlier and not on power saving mode: E000009_ 202A_ 00012400a_ 1-- Valid 8-12-08 09:01:11 10-12-08 03:12:17 9.34 140.2 / 140.2 E000009_ 202A_ 00012400a_ 0-- Valid 8-12-08 09:00:36 9-12-08 08:41:53 15.59 238.2 / 140.2 <Must be an outlier :P [Added words: Removed the Quad from CEP. Too much kernel time and PF/PF delta. Devices in quorum show hours less run time claiming virtually same hourly credit suggesting material inefficiency on quad. sample E000012_ 785A_ 00021z00n_ 0-- Valid 9-12-08 14:39:55 10-12-08 11:16:53 8.58 130.9 / 114.1 < 15.25 phr E000012_ 785A_ 00021z00n_ 1-- Valid 9-12-08 14:39:38 10-12-08 04:50:04 6.31 97.4 / 114.1 < 15.44 phr]
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Dec 10, 2008 12:16:53 PM] |
||
|
|
|