Search found 30 matches

by mrshirts
Mon Jun 27, 2011 8:23 pm
Forum: Issues with a specific WU
Topic: Project: 3866 (Run 2597, Clone 0, Gen 0)
Replies: 17
Views: 3944

Re: Project: 3866 (Run 2597, Clone 0, Gen 0)

After discussions, we've decided to raise up the final deadline time, by about 30%, so that should solve some of the problems on this thread. Are not adjusting the preferred deadline time quite yet, or the points. Still evaluating that!
by mrshirts
Sat Jun 25, 2011 3:36 am
Forum: Issues with a specific WU
Topic: Project: 3866 (Run 3811, Clone 0, Gen 0)
Replies: 4
Views: 882

Re: Project: 3866 (Run 3811, Clone 0, Gen 0)

Thanks for the report. 3867 should be the same as 3866, so that's strange that is behaving so poorly. It's actually at very low weight compared to the others, so I'll take it off until we understand better what is going on.
by mrshirts
Mon Jun 20, 2011 1:59 am
Forum: Issues with a specific WU
Topic: Project: 3867 (Run 1685, Clone 0, Gen 3) Early Unit End
Replies: 5
Views: 973

Re: Project: 3867 (Run 1685, Clone 0, Gen 3) Early Unit End

Thanks very much for all this information. Most of the WU are returning just fine, but there are a small but nonnegligible fraction that are having these unexplained crashes. We're still in the process of trying to replicate and fix this bug -- it's one of those horrible bugs where when you even put...
by mrshirts
Fri Jun 17, 2011 8:42 pm
Forum: Issues with a specific WU
Topic: Project: 3866 (Run 2597, Clone 0, Gen 0)
Replies: 17
Views: 3944

Re: Project: 3866 (Run 2597, Clone 0, Gen 0)

This series does take a long time - from 44 hours for my 2 GHz dual-core laptop to 77 hours for my 3.0 GHz P4 to 110 hours for my 2.0 GHz Sempron. None report SSE but all report "Assembly optimizations on if available." 333 points seems a little low but better than some others. Hmm. It se...
by mrshirts
Thu Jun 16, 2011 6:55 pm
Forum: Issues with a specific WU
Topic: Project: 3866 (Run 2597, Clone 0, Gen 0)
Replies: 17
Views: 3944

Re: Project: 3866 (Run 2597, Clone 0, Gen 0)

I don't think SSE is the issue here, because -forceasm is being used. The line 01:55:49:Unit 01:- Assembly optimizations manually forced on. Indicates that any problem there is being ignored. If it IS the issue deleting the core.dat file should eliminate this; that's the file that stores that inform...
by mrshirts
Thu Jun 16, 2011 5:05 pm
Forum: Issues with a specific WU
Topic: Project: 3866 (Run 2597, Clone 0, Gen 0)
Replies: 17
Views: 3944

Re: Project: 3866 (Run 2597, Clone 0, Gen 0)

Hi, Bruce- Right now, the problem is that it usually returns correctly from a checkpoint. It appears to occasionally fail, but in beta, we couldn't collect enough information to actually figure out when it fails. It might be best to pull it back to advanced, however - that will presumably allow us t...
by mrshirts
Thu Jun 16, 2011 3:40 am
Forum: Issues with a specific WU
Topic: Project: 3866 (Run 2597, Clone 0, Gen 0)
Replies: 17
Views: 3944

Re: Project: 3866 (Run 2597, Clone 0, Gen 0)

So, Client 7, on Windows XP. I don't think we've tested that combination, so that might be the reason. Is this replicatable? Can if you get another WU (near the beginning, hopefully!) and see if it crashes, that would be very valuable!
by mrshirts
Wed Jun 15, 2011 4:20 am
Forum: Issues with a specific WU
Topic: Project: 3866 (Run 2597, Clone 0, Gen 0)
Replies: 17
Views: 3944

Re: Project: 3866 (Run 2597, Clone 0, Gen 0)

By "startup" I mean restarting on checkpoint.
by mrshirts
Wed Jun 15, 2011 4:20 am
Forum: Issues with a specific WU
Topic: Project: 3866 (Run 2597, Clone 0, Gen 0)
Replies: 17
Views: 3944

Re: Project: 3866 (Run 2597, Clone 0, Gen 0)

So, this is dying only on startup, correct?
by mrshirts
Thu Jun 09, 2011 7:38 pm
Forum: New Donors start here
Topic: performance : normal or big ? (dual core)
Replies: 10
Views: 2530

Re: performance : normal or big ? (dual core)

The stats for 3865-3867 are indeed set up to give a bonus. Note that because of some quirks of hard coding, A6 can only be used on SMP using the v7 client (now in beta).
by mrshirts
Tue Jun 09, 2009 11:41 pm
Forum: Issues with a specific server
Topic: 128.143.48.226 power issues
Replies: 3
Views: 1051

Re: 128.143.48.226 power issues

Well, that was quick. As soon as I posted, the power went back on. Server is now restarted and reloading all the jobs, should be fine (unless the storm starts another blackout).
by mrshirts
Tue Jun 09, 2009 11:29 pm
Forum: Issues with a specific server
Topic: 128.143.48.226 power issues
Replies: 3
Views: 1051

128.143.48.226 power issues

There's an ongoing power outage in the building housing the machines 128.143.48.226 and 128.143.48.227 (the collection server for 128.143.48.226), starting at about 5 p.m. today. (2.5 hours ago). They were on battery backup, but not emergency power, so I had to shut them down. Hopefully the power wi...
by mrshirts
Wed May 27, 2009 7:34 pm
Forum: Issues with a specific WU
Topic: Project 3860 - 3861 (Mass error)
Replies: 11
Views: 1925

Re: Project 3860 - 3861 (Mass error)

The scientific core is designed to autodetect the presence of SSE2, and only run SSE2 instructions if they are applicable; this works on the machines I tested with.

Perhaps if the CPU is old enough (not even SSE), then the signal used to detect SSE2 is not available, and it defaults to SSE2?
by mrshirts
Mon Apr 20, 2009 1:35 am
Forum: Issues with a specific server
Topic: 128.143.48.227 In Reject Mode
Replies: 3
Views: 969

Re: 128.143.48.227 In Reject Mode

A error in the new server cropped up in parsing some assignment server instructions. A workaround has been implemented, and the server restarted. It should show up as soon as it processes all the new jobs.
by mrshirts
Tue Apr 07, 2009 4:42 pm
Forum: Issues with a specific server
Topic: 128.143.48.227 ?
Replies: 7
Views: 1824

Re: 128.143.48.227 ?

It was not showing the CS because of the difference in configuration files in the new and old server code. I've put a fix in place so that the CS now gets listed.