a lot of P2682 are BAD - possible solution !

[noorman]

.

they clog up a series of systems running high speed i7 CPU's which all have the same problem; crashing their WU's, all 2682 ...

there 's something really wrong with this WU family !

report from a friend:

"Not one of them has run on any machine I have. They have ALL generated exactly the same error.every
time without fail. They either give me a 2682 with will not run without that error, or the work server refuses
to give us anything else to do and just sits there for hours saying no more work to do. They've choked the
system in faulty p2682's as far as we can tell and will decide later today what course of action to take."

.

[PantherX]

This Thread contains some important information that you may want to view: viewtopic.php?f=55&t=15563

[7im]

You report doesn't state what the errors are, as reported by your friend.

How much memory does your friend's system have?

[noorman]

.

I 've asked for some Log copies ...

.

[noorman]

.

It 's always this:

Code: Select all

[05:15:33] Project: 2682 (Run 8, Clone 17, Gen 18)
[05:15:33]
[05:15:35] Assembly optimizations on if available.
[05:15:35] Entering M.D.
[05:15:51] Gromacs cannot continue further.
[05:15:51] Going to send back what have done -- stepsTotalG=250000
[05:15:51] Work fraction=0.0000 steps=250000.
[05:15:54] CoreStatus = C0000005 (-1073741819)
[05:15:54] Client-core communications error: ERROR 0xc0000005
[05:15:54] Deleting current work unit & continuing...

.

there were several instances of this on several machines at the same time / have no further info (RCG) on those, but that doesn't seem to matter because they 're all sick in the same bed ...

My friend, who supplied me with this Log info, is stopping Windows SMP -bigadv because, as a Pro (UNIX) top programmer he thinks this is too bad of a show from the programmers who released these badly constructed WU's.
He 's generating extra power himself, to run them, but enough is enough / "this kind of Work Units aren't worth the energy costs one bit !"

He 'll keep on running the stable GPU Folding hardware, but will stop SMP Folding because of this mess ...

.

[noorman]

.

Just got in a message which seems to have a very good lead to the cause of the trouble:

I may have found a workaround. Manually setting memory to something just under 6 gigs
in the config file is letting 2682's come in and start up just fine. Now its a waiting game
to see if they keep running and don't bomb out. I'd advise anyone wanting to try this method
to resize their page file size accordingly in the system properties and profile stuff.

.

Same person who was reporting the above ...

.

[noorman]

You report doesn't state what the errors are, as reported by your friend.

How much memory does your friend's system have?

.

As he now stated, 6 GB of system RAM ...

.

[kasson]

This sounds like a memory per core issue. Is your friend running in a VM? What was the memory prior to the change?

(In general, both Pande group members and other forum helpful folk are able to give the best advice if you can provide full details in an initial bug report. As you do here, the best place to report issues with a WU is with a post in the forum. Thanks!)

[sortofageek]

Project: 2682 (Run 8, Clone 17, Gen 18)

Two folders have been unable to complete this one so far. While it is possible that particular one is bad, the evidence isn't clear enough yet to mark it bad.

[noorman]

This sounds like a memory per core issue. Is your friend running in a VM? What was the memory prior to the change?

(In general, both Pande group members and other forum helpful folk are able to give the best advice if you can provide full details in an initial bug report. As you do here, the best place to report issues with a WU is with a post in the forum. Thanks!)

.

No, he is running native Windows, no VM of any kind ...

.

[toTOW]

Check announcement : viewtopic.php?p=154820#p154820

[noorman]

.

It has nothing to do with the (total) memory requirements ...

.