Minirosetta v1.54 Compute Errors

Message boards : Number crunching : Minirosetta v1.54 Compute Errors

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
epcorian

Send message
Joined: 1 Jan 09
Posts: 16
Credit: 253,062
RAC: 0
Message 59312 - Posted: 5 Feb 2009, 3:24:44 UTC

I just noticed last night that both my PC's are getting compute errors with minirosetta, like its bad, real bad, my apologies for all the failed WU's. I usually crunch SETI with no problems and just the other day I decided to crunch Rosetta part time so this all new to me. I've noticed that my Beta's don't fail, but the Mini's fail almost instantly on both PC's. I keep getting an Exit status of -1073741819 (0xc0000005), you can see the full output on Workunit 206160079 if you look under the Task ID 226196786.

So to fix this so far on my Q6600 PC I have:
Updated my bios.
Updated my motherboard drivers in Windows.
Updated my graphics driver.
Updated my DirectX 10.
Made sure graphics are turned off.
Uninstalled and reinstalled BOINC x64 as I am running Vista 64.
Reset the Rosetta project.
Now I am just waiting for memtest86+ to finish, as we speak 40% done, no errors yet.

Then I will try and hopefully crunch a Mini unit successfully. Oh and I do have it set to leave applications in memory when switching, I have 4GB of RAM in my Q6600 PC.

Is anything else I am missing?

Thanks,
E
ID: 59312 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
epcorian

Send message
Joined: 1 Jan 09
Posts: 16
Credit: 253,062
RAC: 0
Message 59313 - Posted: 5 Feb 2009, 4:03:00 UTC

Me again, memtest passed with no errors, so I fired up BOINC and it downloaded 5 Minirosetta v1.54 work units and they all failed in less then a minute...the same as they did before.

Any ideas?? I don't want to give up on Rosetta as I think it actually makes a difference.
ID: 59313 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Paul D. Buck

Send message
Joined: 17 Sep 05
Posts: 815
Credit: 1,812,737
RAC: 0
Message 59317 - Posted: 5 Feb 2009, 5:56:28 UTC

Is anything else I am missing?


Did you try turning off OC?

Different applications react differently on systems. Your system can be fine for SaH or even Memtest ... but not for Rosetta ...

ID: 59317 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Greg_BE
Avatar

Send message
Joined: 30 May 06
Posts: 4877
Credit: 4,586,538
RAC: 337
Message 59323 - Posted: 5 Feb 2009, 10:05:06 UTC - in response to Message 59317.  

Is anything else I am missing?


Did you try turning off OC?

Different applications react differently on systems. Your system can be fine for SaH or even Memtest ... but not for Rosetta ...




I know this to be a fact as well. I had to much OC and bombed tasks right from the start or they would almost finish and then crash. Einstein had no trouble with my large OC speed but Rosie didn't like it one bit.
ID: 59323 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
LizzieBarry

Send message
Joined: 25 Feb 08
Posts: 76
Credit: 201,862
RAC: 0
Message 59325 - Posted: 5 Feb 2009, 11:22:38 UTC

Also, ensure the setting of "On multiprocessor systems, use at most ___% of the processors" to 100% in BAM's "AdvancedPreferences|processor usage" tab seems to be a factor.
ID: 59325 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
epcorian

Send message
Joined: 1 Jan 09
Posts: 16
Credit: 253,062
RAC: 0
Message 59338 - Posted: 5 Feb 2009, 15:09:06 UTC

Hey guys, thanks for help.

I'm not OC'ing my Q6600, everything is running at stock speeds.

But I do have my processor throttled to 75% so that it only uses 3 of the 4 cores for BOINC projects, I will see if changing it to 100% makes a difference, thanks for the tip.

ID: 59338 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 59339 - Posted: 5 Feb 2009, 15:50:18 UTC

epcorian as you learn more about your outcome, please post to the stickied v1.54 thread. There seems to be some cases where less then 100% has some intermittant problems. Did you configure to use 3CPUs? or a max of 75%? Sounds like the 75%, so did the BOINC Manager actually show 4 tasks in a running status at the same time? And then each of the 4 is suspended briefly to enforce the 75%?

If you could post answers to the above along with what you find when you run at 100% to the other thread, that may help track down the issue.
Rosetta Moderator: Mod.Sense
ID: 59339 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
epcorian

Send message
Joined: 1 Jan 09
Posts: 16
Credit: 253,062
RAC: 0
Message 59341 - Posted: 5 Feb 2009, 16:15:33 UTC - in response to Message 59325.  

Also, ensure the setting of "On multiprocessor systems, use at most ___% of the processors" to 100% in BAM's "AdvancedPreferences|processor usage" tab seems to be a factor.


Mine is set like this right now: "On multiprocessor systems, use at most 75% of the processors"

Is there a different way of telling it to use just only 3 of 4 cores instead of setting it to 75%?

Regardless, I will change it to use 100% when I get home from work and post back here and in the debug thread.

If it's possible to get working, I would like to only use 3 of 4 cores but I'll worry about that later once we figure out what is going on.

Thanks!
ID: 59341 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
epcorian

Send message
Joined: 1 Jan 09
Posts: 16
Credit: 253,062
RAC: 0
Message 59344 - Posted: 5 Feb 2009, 16:26:16 UTC

Oh and to answer your question, yes, if I had 4 tasks, 3 would be running and the 4th would be sitting there saying "Waiting to run" or "Ready to start"
ID: 59344 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
epcorian

Send message
Joined: 1 Jan 09
Posts: 16
Credit: 253,062
RAC: 0
Message 59370 - Posted: 6 Feb 2009, 1:11:35 UTC

I changed it to: "On multiprocessor systems, use at most 100% of the processors" so that it would run completely unthrottled and use all 4 cores. And I let it download minirosetta WU's and it got 5 of them and all failed after 0:33, 1:39, 0:56, 0:38, and last one at 0:51 crashed with a Vista popup saying "minirosetta_1.54_windows_x86_64.exe has stopped working"

I don't know what else to do to get minirosetta working but I have to say that I'm ashamed of looking at my results screen and seeing compute error all over the place! I don't think its fair to this project and all the successful crunchers to keep piling up errors! I seem to be able to crunch Beta WU's okay as long as I don't start messing with settings and starting/stopping the beta crunching so maybe I will just stick with those.

I think Vista 64 is most likely my culprit...

Thanks for the help everyone!
ID: 59370 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
transient
Avatar

Send message
Joined: 30 Sep 06
Posts: 376
Credit: 10,836,395
RAC: 84
Message 59380 - Posted: 6 Feb 2009, 6:30:47 UTC

You could utilize the "use at most x CPU's" setting to limit BOINC to 3 cores in stead of 4.
ID: 59380 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
voyager
Avatar

Send message
Joined: 2 Feb 08
Posts: 5
Credit: 76,584
RAC: 0
Message 59381 - Posted: 6 Feb 2009, 8:52:57 UTC
Last modified: 6 Feb 2009, 8:53:27 UTC

You could save this as cc_config.xml and place in boinc file.

<cc_config>

<options>
<ncpus>3</ncpus>
</options>
</cc_config>
ID: 59381 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mikey
Avatar

Send message
Joined: 5 Jan 06
Posts: 1828
Credit: 5,957,967
RAC: 160
Message 59387 - Posted: 6 Feb 2009, 12:40:46 UTC - in response to Message 59370.  

I think Vista 64 is most likely my culprit...
Thanks for the help everyone!


I replied to Mod Sense in the other thread about this and hopefully you will get your answer. What I said was that I read somewhere that people with 64 bit OS's had to do something to make the 32 bit workunits work and I was wondering if Rosetta was one of those projects.
ID: 59387 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
epcorian

Send message
Joined: 1 Jan 09
Posts: 16
Credit: 253,062
RAC: 0
Message 59408 - Posted: 7 Feb 2009, 3:37:36 UTC

Thanks guys, I think I've got it working utilizing all 4 cores, but it's still early, I am just crunching my first WU of 8 Mini's...so far it has run 30 minutes without crashing and burning! Much better then the typical 30-60 seconds I was getting before.

My solution, downgrade to the 6.2.19 64-bit of the BOINC client. I'll let you know if I can successfully crunch a mini!

Then I will see if I can restrict to 3 of 4 cores and still successfully crunch.
ID: 59408 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 59438 - Posted: 8 Feb 2009, 2:30:55 UTC - in response to Message 59387.  

I think Vista 64 is most likely my culprit...
Thanks for the help everyone!


I replied to Mod Sense in the other thread about this and hopefully you will get your answer. What I said was that I read somewhere that people with 64 bit OS's had to do something to make the 32 bit workunits work and I was wondering if Rosetta was one of those projects.


No, there's nothing you can adjust on the work units. I think what you are thinking of is projects that do not recognize a 64bit OS at all and will not send work.
Rosetta Moderator: Mod.Sense
ID: 59438 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 1636
Credit: 29,396,895
RAC: 16,810
Message 59480 - Posted: 9 Feb 2009, 14:04:11 UTC - in response to Message 59341.  

Mine is set like this right now: "On multiprocessor systems, use at most 75% of the processors"

Is there a different way of telling it to use just only 3 of 4 cores instead of setting it to 75%?

Regardless, I will change it to use 100% when I get home from work and post back here and in the debug thread.

If it's possible to get working, I would like to only use 3 of 4 cores but I'll worry about that later once we figure out what is going on.

Thanks!

I run a quad core with Vista 64 and used to have the machine throttled to 50%. It still utilised 4 cores but turned them on and off to run each 50% of the time. It didn't turn any cores off. I checked this with a CPU sidebar gadget.

The <ncpus>3</ncpus> option in cc_config.xml sounds a more likely solution, though I admit I've never used it.
ID: 59480 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Paul D. Buck

Send message
Joined: 17 Sep 05
Posts: 815
Credit: 1,812,737
RAC: 0
Message 59487 - Posted: 9 Feb 2009, 18:42:20 UTC

Rosetta is one of those projects where the tasks do not seem to take kindly to the use of CPU throttling ... I know when I was using a setting in BOINC to limit to 99% I was having a 40% failure rate.

This is a topic of interest to the testing team, so if you are using CPU throttling, and are having failures, and those failures go away if you turn off the throttle ... well, all I can say is that it worked for me ...
ID: 59487 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 59491 - Posted: 9 Feb 2009, 21:33:25 UTC

Sid, the ncpus option in the cc_config is intended for test cases where you want to behave as though you have more CPUs then your machine has. For example, allowing a dual core to run 5 tasks at a time. They each just compete for CPU (and memory).

The reccommendation though if you wish to use less then all of the CPUs it to make such a setting in the general preferences on the website, or in the local preferences for each machine.
Rosetta Moderator: Mod.Sense
ID: 59491 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Paul D. Buck

Send message
Joined: 17 Sep 05
Posts: 815
Credit: 1,812,737
RAC: 0
Message 59496 - Posted: 10 Feb 2009, 0:45:20 UTC - in response to Message 59491.  

Sid, the ncpus option in the cc_config is intended for test cases where you want to behave as though you have more CPUs then your machine has. For example, allowing a dual core to run 5 tasks at a time. They each just compete for CPU (and memory).

The reccommendation though if you wish to use less then all of the CPUs it to make such a setting in the general preferences on the website, or in the local preferences for each machine.


It also is used to allow the use of CUDA with some versions of BOINC Manager below 6.5.0 for which it is not needed.
ID: 59496 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
epcorian

Send message
Joined: 1 Jan 09
Posts: 16
Credit: 253,062
RAC: 0
Message 59611 - Posted: 16 Feb 2009, 16:42:23 UTC

So this weekend I installed a fresh copy of XP x64, upgraded it to SP2, installed my x64 version of NOD32 antivirus, told BOINC to use "...use at most 75% of the processors" meaning 3 of 4 cores on my Q6600 and it's crunching Mini's and Beta's without a problem! 1 successful Beta, 5 successful Mini's with 4 more coming down the pipe. So it looks like Mini does not like Vista x64 and on my adventures on google, it turns out that XP x64 is actually based on the Server 2003 code tree while Vista is based on crap. :)
ID: 59611 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
1 · 2 · Next

Message boards : Number crunching : Minirosetta v1.54 Compute Errors



©2021 University of Washington
https://www.bakerlab.org