Advanced search

Message boards : Multicore CPUs : Reporting bad test work unit

Author Message
Jesse Viviano
Send message
Joined: 9 Jan 15
Posts: 8
Credit: 49,960,837
RAC: 183,663
Level
Val
Scientific publications
wat
Message 39480 - Posted: 14 Jan 2015 | 19:47:15 UTC
Last modified: 14 Jan 2015 | 19:47:41 UTC

Work unit 10518380 is apparently bad, at least on the 32-bit Windows multi-threaded SSE2 application when running on 64-bit Windows 7 and a Core i7-980X. It consistently caused my machine to crash, forcing me to abort it.

I am aware of the need for running test applications so that things can get debugged, but a work unit that consistently crashes and forces me to abort it is not going to help you at all unless you have useful debug data. I do not know if the unhandled exception is caused by an abort or by the program crashing, so I am reporting the work unit here so that the developers know to investigate the crash.

Jesse Viviano
Send message
Joined: 9 Jan 15
Posts: 8
Credit: 49,960,837
RAC: 183,663
Level
Val
Scientific publications
wat
Message 39492 - Posted: 16 Jan 2015 | 4:27:39 UTC

Here is another bad test work unit: http://www.gpugrid.net/workunit.php?wuid=10518376. The circumstances were the same with this one as the previous bad one. Between the two bad work units, I was able to complete one somewhat good test work unit that caused one crash but recovered successfully and completed, http://www.gpugrid.net/workunit.php?wuid=10518302

Jesse Viviano
Send message
Joined: 9 Jan 15
Posts: 8
Credit: 49,960,837
RAC: 183,663
Level
Val
Scientific publications
wat
Message 39502 - Posted: 16 Jan 2015 | 17:54:17 UTC

One of the things I do with my computer due to family requirements is that I shut it down each night and have to interrupt jobs in progress and force them to restore from checkpoints when the computer is turned on in the morning. Could this be a reason the work units fail?

Jesse Viviano
Send message
Joined: 9 Jan 15
Posts: 8
Credit: 49,960,837
RAC: 183,663
Level
Val
Scientific publications
wat
Message 39503 - Posted: 16 Jan 2015 | 22:01:04 UTC

The crashes could have been caused by a CPU that needed reseating, so the work units might not be bad after all. I will post what I find with the last work unit I am crunching.

Jesse Viviano
Send message
Joined: 9 Jan 15
Posts: 8
Credit: 49,960,837
RAC: 183,663
Level
Val
Scientific publications
wat
Message 39506 - Posted: 17 Jan 2015 | 3:30:01 UTC

My final test work unit finished after a few crashes. I have quit doing beta work units after this because I do not know if the problem is my CPU or bugs in the test program and therefore my bug reports may be suspect. I might go back if I get the funds to build a new computer (mine has a 5 year old Core i7 980X). I have not had these types of crashes before I joined the beta work unit program. (I am calling it a beta work unit program instead of a test work unit program because the executable is in a beta state rather than the data and parameters being run through the program that need testing.)

Post to thread

Message boards : Multicore CPUs : Reporting bad test work unit