Page 1 of 1

A question

PostPosted: Mon Jun 08, 2020 8:07 am
by LeonardRenner
This week I upgraded my GPU. The old GPU, an HD5550, had about a day left to complete its WU, and I estimated that I could swap in the new one and, even if there was a problem and I had to roll back, there would still be enough time for the old GPU to complete the WU.

This didn't happen though. When the new card was installed, FAH client just discarded the old WU and got a new one.

Is there anyway to avoid this, other than simply waiting for the current WU to complete?

Also, is this really desirable default behaviour from the client? Surely the new GPU should just finish the current WU?

Re: A question

PostPosted: Mon Jun 08, 2020 8:42 am
by PantherX
Welcome to the F@H Forum LeonardRenner,

I am not sure what model your new GPU was. However, for AMD, there are Projects that are limited to one set of hardware than the other simply because there's a technical limitation where Navi doesn't run FahCore_21 while non-Navi GPUs can run FahCore_21. Thus, if the WU was FahCore_21 and you upgraded to a Navi GPU, then dumping the WU is expected.

Generally speaking, it is recommended to finish folding WUs (use the Finish option in Advanced Control) when upgrading hardware. If in doubt, ask in the Forum or finish :)

Re: A question

PostPosted: Mon Jun 08, 2020 2:06 pm
by Joe_H
Which WU were you folding on the HD 5550? That GPU does not support Double Precision (FP64) and was removed from the supported list a while ago.

But in any case, while it might be desirable that a WU continue on with a different GPU, architectural changes between generations usually mean there is a chance that the calculations will be different.

Re: A question

PostPosted: Mon Jun 08, 2020 3:00 pm
by JimboPalmer
Welcome to Folding@Home!

Your HD 5550 is a card in the Terascale family, they used instructions called VLIW ... chitecture)#TeraScale_2 ... ction_word

After 3 generations of Terascale, AMD used GCN Instructions for 5 generations.

Most recently, AMD has moved to another instruction set named RDNA.

None of these are interchangeable, and none are interchangeable with the various Nvidia or Intel GPUs.