Amd Vulkan beta driver & Fah

It seems that a lot of GPU problems revolve around specific versions of drivers. Though AMD has their own support structure, you can often learn from information reported by others who fold.

Moderators: Site Moderators, PandeGroup

Re: Amd Vulkan beta driver & Fah

Postby DarkFoss » Fri Apr 22, 2016 6:07 pm

bruce wrote:
DarkFoss wrote:... gpu load never hits 100% always stops at 99% temps low too 48-49c...


I wouldn't worry about that 1%. There's always some time spend moving data between RAM and VRAM through the bus. The time to move data actually causes a bigger percentage of wait-state on a fast card than on a slower one because the computations are faster while the data transfers don't change. The GPU will run out of work sooner until more can be supplied.

What's your PCIe bus speed? ... and as toTOW suggests, are you sure your CPU can keep up with the demand?


Thanks Bruce, Your explanation does help clarify what I'm seeing. My cpu is fine FX8350 it can handle gaming even when I hook it up to my 4k TV and fold on the secondary Fury X it only impacts FAH's TPF by several seconds. Both card are operating at Pci-E 2.0 x16.

I did not mean to imply that WU 13104 is bad they have all completed with a TPF of 1.57 to 2.00 min. Perhaps this WU is just a series of quick operations and the card doesn't need to hit 100%. I'd just say anyone that has a Fury X Fury or Nano don't be quick to judge a new driver on this wu unless the TPF is way off and if it is it could just be a case on an install gone sideways.
Image
DarkFoss
 
Posts: 70
Joined: Fri Apr 16, 2010 11:43 pm
Location: DG,IL

Re: Amd Vulkan beta driver & Fah

Postby DarkFoss » Fri Apr 22, 2016 6:40 pm

bruce wrote:I'm talking to anyone who is complaining that their GPU isn't showing 100%. It might be drivers or it might be hardware or it might be an OpenCL limitation that needs attention.

For what it's worth I have screenshots of cpu-z sensor tab along with the corresponding cpu-z log files for wu's 13103 which shows the same behavior as 13014. 10468 and 10495 which show 100% gpu load all taken yesterday. I could share them on Onedrive or upload to Google docs if there is any interest, captured with the default 1.0 second so not to induce any lag.
*Just like the 13104, I am not suggesting there is anything wrong with the 13103( average TPF of 1:57) just recommending not to judge any new driver solely on these wu's.

They all are with the new bios AMD released on Monday for Fury X and Nano. Note these are NOT for the Fury non X.
http://support.amd.com/en-us/download/g ... e-download
DarkFoss
 
Posts: 70
Joined: Fri Apr 16, 2010 11:43 pm
Location: DG,IL

Re: Amd Vulkan beta driver & Fah

Postby Bryman » Fri Apr 22, 2016 10:49 pm

DarkFoss wrote:
bruce wrote:
DarkFoss wrote:... gpu load never hits 100% always stops at 99% temps low too 48-49c...


I wouldn't worry about that 1%. There's always some time spend moving data between RAM and VRAM through the bus. The time to move data actually causes a bigger percentage of wait-state on a fast card than on a slower one because the computations are faster while the data transfers don't change. The GPU will run out of work sooner until more can be supplied.

What's your PCIe bus speed? ... and as toTOW suggests, are you sure your CPU can keep up with the demand?


Thanks Bruce, Your explanation does help clarify what I'm seeing. My cpu is fine FX8350 it can handle gaming even when I hook it up to my 4k TV and fold on the secondary Fury X it only impacts FAH's TPF by several seconds. Both card are operating at Pci-E 2.0 x16.

I did not mean to imply that WU 13104 is bad they have all completed with a TPF of 1.57 to 2.00 min. Perhaps this WU is just a series of quick operations and the card doesn't need to hit 100%. I'd just say anyone that has a Fury X Fury or Nano don't be quick to judge a new driver on this wu unless the TPF is way off and if it is it could just be a case on an install gone sideways.


13000 projects require more cpu than the other ones, set it so 2 full cores are free and see if that helps

For my i7 if I had it set to use 7 out of 8 threads then the GPU usage would be about 30%, changed that to 6 out of 8 threads and GPU usage went up to 95%+


Edit: Nevermind just read the whole post... your GPU usage is normal, most of the time it'll be 97-98% no matter what WU it is


The usage is an APPROXIMATION... when I'm gaming and I have v-sync turned off, it'll stay at 100% the whole time... but my temps will vary depending on the camera angle... so that means that the usage is increasing at times but it purposely shows it hit 100% before it actually hits 100%

Depending on what calculations the GPU needs to make, it'll hit more or less usage.. but it should still be running at 100%, as long as you have a core or 2 threads free on the CPU... maybe the CPU will drop the usage by 1%, but that's it

So back to OCCT... temps skyrocket, that's because the video card is able to put out near 100% usage unlike games which are like 60% and folding@home which is like 50%

GPU usage isn't as accurate as CPU usage

Well... technically OCCT isn't a good example...

nevermind...

Point is that folding@home does a lot of certain calculations that slow the card down, it can't be helped
Bryman
 
Posts: 95
Joined: Mon Mar 14, 2016 12:01 pm

Re: Amd Vulkan beta driver & Fah

Postby Sven » Mon Apr 25, 2016 1:33 pm

There are very big PPD differences with the AMD Fury X.

At the moment I got one of the 13103 with a PPD of 290'000. Before that I got a 9206 with a PPD I've never seen on the Fury X: 560'000 . When I remember correctly, the 9206 is one with a lot of Atoms. Seems as if the Fury X likes those.
Sven
 
Posts: 33
Joined: Fri Nov 01, 2013 8:12 pm

Previous

Return to Problems with AMD/ATI drivers

Who is online

Users browsing this forum: No registered users and 2 guests

cron