I have been doing tasks using Intel HD 630 GPU, they were consistently running around 80 Minutes, they now take around 4 Hours , has anything changed that would cause this, CPU is an Intel I7 . Wouldn’t mind but the credits are the same.
Along this same idea of the FGRP#1 gpu tasks does anyone have a good idea of how much memory it takes to run more than one task at a time or is it all hit or miss? I have gpu's crunching FGRP#1 tasks that have anywhere from 2gb of onboard memory to one with 12gb of onboard memory and cards with 4gb, 6gb and 8gb onboard memory and am wondering where it makes more sense to run multiple tasks instead of a single task at a time. I am not really interested in pushing cards to their limit, I want a good long life out of them but still if a 2gb card can do a task surely a 6gb or even 8gb gpu could do 2 at a time...couldn't it?
On a 2080 card with 8GB of VRAM, I can do two tasks at a time and consume 6525MB of VRAM between the two tasks. So yes it is possible to run two tasks per card.
As long as the time for a reported task running two at a time is less than half the reported total time for running singles, it is beneficial to run doubles.
Also try and stagger the start-finish of the doubled tasks so that neither of them occupy the last 90-100% of the completion stage at the same time since that slows down both of them.
On a 2080 card with 8GB of VRAM, I can do two tasks at a time and consume 6525MB of VRAM between the two tasks. So yes it is possible to run two tasks per card.
As long as the time for a reported task running two at a time is less than half the reported total time for running singles, it is beneficial to run doubles.
Also try and stagger the start-finish of the doubled tasks so that neither of them occupy the last 90-100% of the completion stage at the same time since that slows down both of them.
Thank you very much!! I did the stagger start thing, on my 3060, and it seems to working just fine not to work on the other gpu's with 8gb of ram on them.
I have been doing tasks using
)
I have been doing tasks using Intel HD 630 GPU, they were consistently running around 80 Minutes, they now take around 4 Hours , has anything changed that would cause this, CPU is an Intel I7 . Wouldn’t mind but the credits are the same.
Along this same idea of the
)
Along this same idea of the FGRP#1 gpu tasks does anyone have a good idea of how much memory it takes to run more than one task at a time or is it all hit or miss? I have gpu's crunching FGRP#1 tasks that have anywhere from 2gb of onboard memory to one with 12gb of onboard memory and cards with 4gb, 6gb and 8gb onboard memory and am wondering where it makes more sense to run multiple tasks instead of a single task at a time. I am not really interested in pushing cards to their limit, I want a good long life out of them but still if a 2gb card can do a task surely a 6gb or even 8gb gpu could do 2 at a time...couldn't it?
On a 2080 card with 8GB of
)
On a 2080 card with 8GB of VRAM, I can do two tasks at a time and consume 6525MB of VRAM between the two tasks. So yes it is possible to run two tasks per card.
As long as the time for a reported task running two at a time is less than half the reported total time for running singles, it is beneficial to run doubles.
Also try and stagger the start-finish of the doubled tasks so that neither of them occupy the last 90-100% of the completion stage at the same time since that slows down both of them.
Keith Myers wrote: On a 2080
)
Thank you very much!! I did the stagger start thing, on my 3060, and it seems to working just fine not to work on the other gpu's with 8gb of ram on them.