Forum > Windows

Computation Error with CUDA

<< < (3/4) > >>

Richard Haselgrove:
Whatever the exit code is, it'll be in there somewhere.

I knew I'd had a -12 on one machine since the servers went off, so those entries were a direct copy from client_state.

Others may appear slightly differently, but <exit_status>xxx</exit_status> will always be a key entry.

Geek@Play:
During past outages I have successfully search "client_state.xml" for the word "error".  On those searches I successsfully located -12 errors and -177 errors.

My first reaction is that they are -177 since they all occured at the same instant during processing.  That was also my experience with the -177 errors.

Frizz:

--- Quote from: Geek@Play on 10 Aug 2010, 06:53:11 pm ---My first reaction is that they are -177 since they all occured at the same instant during processing.  That was also my experience with the -177 errors.

--- End quote ---

Damn it - just found some "Computation Error" for CPU units too. 5 WUs. All at the EXACT same time.
Is this an indication for -177 errors?

By the way: What exactly is this -177 error?

Richard Haselgrove:

--- Quote from: Geek@Play on 10 Aug 2010, 06:53:11 pm ---My first reaction is that they are -177 since they all occured at the same instant during processing.  That was also my experience with the -177 errors.

--- End quote ---


--- Quote from: Frizz23 on 10 Aug 2010, 07:20:38 pm ---Damn it - just found some "Computation Error" for CPU units too. 5 WUs. All at the EXACT same time.
Is this an indication for -177 errors?

By the way: What exactly is this -177 error?

--- End quote ---
You need to be certain exactly what they are - search client_state, don't just rely on our speculation.

-177 errors arise from bad estimates issued by the server - they don't indicate a problem at your end. At this moment, I would guess they're re-issues for the ghost WUs which are timing out - look for _2, _3 or higher on the end of the task name. Don't worry about them unless you get a huge number - if you do, Fred's latest rescheduling tool can deal with them, by resetting the estimates.

Frizz:

--- Quote from: Richard Haselgrove on 10 Aug 2010, 07:27:04 pm ---You need to be certain exactly what they are - search client_state, don't just rely on our speculation.

--- End quote ---

OK found it (-177). It would be nice if the BOINC client would show this - and not just "Computation error". Especially at times when SETI is offline.

How can I prevent this -177 error?

Navigation

[0] Message Index

[#] Next page

[*] Previous page

Go to full version