Question about AMD GPU and checkpoints

If you think it might be a driver problem, see viewforum.php?f=79

Moderators: Site Moderators, FAHC Science Team

Joe_H
Site Admin
Posts: 7937
Joined: Tue Apr 21, 2009 4:41 pm
Hardware configuration: Mac Pro 2.8 quad 12 GB smp4
MacBook Pro 2.9 i7 8 GB smp2
Location: W. MA

Re: Question about AMD GPU and checkpoints

Post by Joe_H »

Thanks for posting that McAfee was the problem. The inability to exclude the entire directory is definitely a non-starter for use with F@h and many other software applications.
mwroggenbuck wrote:I did find something a little strange. I cannot put my computer into sleep mode unless I pause and exit FAH. Not a big deal, but it was funny to see my computer stay alive when I told it to sleep (screens would go out, but computer would not stop)
This is actually an intentional setting in the client. A computer going to sleep while GPU folding is still in progress will corrupt the WU. This happens because the contents of the memory and registers on the GPU are not saved when sleep mode is entered. So when the computer comes out of sleep, the state of the GPU contents is not what the GPU folding core running on the CPU left it as.

They have made improvements in the code to better detect this happening, and have the core stop and go back to the last checkpoint. But that is not something that works 100% of the time.

If you were doing folding on just the CPU this is not an issue. During sleep mode contents of RAM are either maintained or in the case of the various deep sleep / hibernation saved to disk. So coming out of sleep the RAM contents will be consistent with what the CPU core expects.

There is a setting that can be used to override this default, but you would still need to pause GPU folding first.
Image

iMac 2.8 i7 12 GB smp8, Mac Pro 2.8 quad 12 GB smp6
MacBook Pro 2.9 i7 8 GB smp3
mwroggenbuck
Posts: 127
Joined: Tue Mar 24, 2020 12:47 pm

Re: Question about AMD GPU and checkpoints

Post by mwroggenbuck »

Thanks for the information. Like I said, it is no big deal. I was just kind of surprised that a pause was not enough. It sounds like there is a setting to override that, but I am not interested in it.

I understand about the suspension to memory for CPU but not GPU, but I thought I would mention this side effect in case someone else had it.

I am still folding away just fine. The only loose thread is why was this more apparent with a checkpoint recovery. For me, it happened all the time with a recovery. It did happen once without a recovery, but the previous WU did have a recovery and I figured the flaw just carried over.

It is not real important, but the engineer in me would like to figure it out so I can wrap a bow on this problem. :egeek:
Post Reply