| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Beta Testing
Forum: Beta Test Support Forum
Thread: CEP2 beta for windows - Version 6.25 |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 311
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
I noticed that the CEP2 Beta's jump to 100% complete rather quickly at the end. I have my client set to .02 days Additional Work Buffer and Connect About Every .02 days. It usually downloads new work when there's about 25 min. left on one of the 2 running WU's. With the CEP2 Beta's, they jump to 100% done and finish before triggering the download for new work. The Beta WU that just jumped didn't skip any steps at the end. It ran all the way thru Job 15 with Job 15 taking about 2.5 hrs clock time. Total CPU time was 6.4 hrs. Are the CEP2 WU's going to continue to jump at the end when they switch to production mode for Windows? If so, then I'll have to increase my settings to download sooner. If the profile for the device is set to effectively hope on some Beta work only, you will have idling cores and won't get a new beta until you've reported the last beta as completed (rule: 1 per core on client max). In production, work is provided according cache settings and when production work is available.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
p3nguin53
Advanced Cruncher USA Joined: Dec 8, 2008 Post Count: 95 Status: Offline Project Badges: 
|
If the profile for the device is set to effectively hope on some Beta work only, you will have idling cores and won't get a new beta until you've reported the last beta as completed (rule: 1 per core on client max). In production, work is provided according cache settings and when production work is available. Yes, I set them low to help grab Beta's and then babysit it to report completed Beta's in order to keep the cores busy when running Beta's. But in this case even with other projects selected, it didn't request new work until after the Beta finished. Relevant msgs are below. I only sat idle for 8 secs. But by not downloading 25 mins early to build my buffer, there is a greater risk of getting caught in a server outage with no chance to retry before going idle. Right now I'm just trying to figure out if CEP2 will work like this in production so I can adjust my settings when it goes live. Thanks for the feedback.  9/21/2010 12:02:38 AM|World Community Grid|Finished upload of c4cw_target02_011892797_0_0 9/21/2010 3:09:39 AM|World Community Grid|Computation for task BETA_E200364_857_A.24.C19H11NOS3.62.3.set1d06_2 finished 9/21/2010 3:09:40 AM|World Community Grid|Sending scheduler request: To fetch work. Requesting 3456 seconds of work, reporting 1 completed tasks 9/21/2010 3:09:41 AM|World Community Grid|Started upload of BETA_E200364_857_A.24.C19H11NOS3.62.3.set1d06_2_0 9/21/2010 3:09:41 AM|World Community Grid|Started upload of BETA_E200364_857_A.24.C19H11NOS3.62.3.set1d06_2_1 9/21/2010 3:09:43 AM|World Community Grid|Finished upload of BETA_E200364_857_A.24.C19H11NOS3.62.3.set1d06_2_0 9/21/2010 3:09:43 AM|World Community Grid|Started upload of BETA_E200364_857_A.24.C19H11NOS3.62.3.set1d06_2_2 9/21/2010 3:09:45 AM|World Community Grid|Scheduler request succeeded: got 1 new tasks 9/21/2010 3:09:47 AM|World Community Grid|Finished upload of BETA_E200364_857_A.24.C19H11NOS3.62.3.set1d06_2_1 9/21/2010 3:09:47 AM|World Community Grid|Started upload of BETA_E200364_857_A.24.C19H11NOS3.62.3.set1d06_2_3 9/21/2010 3:09:47 AM|World Community Grid|Starting c4cw_target02_012215721_0 |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
There are observations of clients not going out to get work until a task has finished such as when a client runs in high priority mode or the 'on_frac' stats are screwed up and I think combining with GPU crunching. In short, a client should not but as it is the scheduler has it's quirks. For instance, I had a client on Windows last week after running that device for 5 weeks on Linux. After booting into Windows the client would not fetch work at all until knobbing under the hood (6.10.58)... ludachris. But to underline, CEP2 to has no special fetch rules.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
gb009761
Master Cruncher Scotland Joined: Apr 6, 2005 Post Count: 2977 Status: Offline Project Badges: 
|
I've just picked up 2 NEW BETA WU's - is this an additional test?
----------------------------------------BETA_ E200495_ 970_ A.24.C18H10S4SeSi.23.2.set1d06_ 0-- BETA_ E200495_ 888_ A.24.C22H16Si2.53.3.set1d06_ 0--  |
||
|
|
David_L6
Senior Cruncher USA Joined: Aug 24, 2006 Post Count: 296 Status: Offline Project Badges: 
|
I've just picked up 2 NEW BETA WU's - is this an additional test? I got a few myself. Wondering the same thing....   |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Still version 6.25? If so, not a new science app.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
gb009761
Master Cruncher Scotland Joined: Apr 6, 2005 Post Count: 2977 Status: Offline Project Badges:
|
Yes, it's still 6.25 - just thought that all the WU's had already been sent out (and the vast majority, returned)
---------------------------------------- |
||
|
|
Dataman
Ace Cruncher Joined: Nov 16, 2004 Post Count: 4865 Status: Offline Project Badges: 
|
I came in to find I received 4 new beta's. They are new and not retreads.
----------------------------------------One came in on a Pentium D I was backing up (BOINC was not supposed to be running). This one should run for the next hundred years.    |
||
|
|
David_L6
Senior Cruncher USA Joined: Aug 24, 2006 Post Count: 296 Status: Offline Project Badges:
|
One came in on a Pentium D I was backing up (BOINC was not supposed to be running). This one should run for the next hundred years. I have one on a 3.4GHz P4 Northwood. That one should take a while... |
||
|
|
gb077492
Advanced Cruncher Joined: Dec 24, 2004 Post Count: 96 Status: Offline |
These are new WUs, not repairs. However, I noticed two things:
1) I only got one WU per TWO cores, not one each as usual. 2) A machine which had been running DDDT2 until they ran out, and was then picking up C4CW, was told it had too low a download bandwidth. I had to give it 2 HPF2 tasks to get the bandwidth measure back up so that it would pick up a beta WU again. There is clearly going to be a problem in production due to the very small size of the C4CW WUs and the way that download bandwidth is measured. I can see there being some irate users as a result. It seems to me that the bandwidth measurement process needs a tweak to cope with the small C4CW WU size. The only other alternatives I see are to make the C4CW WUs larger (with dummy data?) to ensure that realistic measurements are taken (which is wasteful) or to lower the CEP2 bandwidth limit and leave it to people to opt out if it's a problem for them (not so good for the set it and forget it brigade). |
||
|
|
|