New AMD GPU Performance To Be Boosted By Linux 4.5; How It Compares To The Binary Blob

[phoronix]

Phoronix: New AMD GPU Performance To Be Boosted By Linux 4.5; How It Compares To The Binary Blob

Last week I posted some AMD proprietary vs. open-source AMD Linux driver benchmarks using the very latest code. Left out of that earlier comparison was the R9 Fury series with Fiji GPU as well as newer graphics cards using the Tonga GPU. These graphics cards are supported by the AMDGPU DRM driver rather than the long-standing Radeon DRM driver. As I've been mentioning a lot this week, Linux 4.5 will bring the PowerPlay power management / re-clocking support to AMDGPU. In this article are showing benchmarks of the Fiji and Tonga GPUs under Linux 4.4 and Linux 4.5 DRM-Next along with the Catalyst 15.9 driver as shipped by Ubuntu 15.10.

http://www.phoronix.com/vr.php?view=22605

[xxmitsu]

Looking for fury tests, I've also found this result file on openbenchmarking http://openbenchmarking.org/result/1...HA-BPPADOKAN90 that shows a beautiful 221.97 fps on 3440 x 1440 with DRI3. What's the explanation between those differences with Michael's result (on a lower 1920 x 1080), then ? could there be a regression somewhere in between ?

[jf33]

In earlier tests, you mentioned that the Fury card only advertised OpenGL 3.3, IIRC. Is that still true? Also, repeating my question from another thread: Does ZeroCore power saving work with AMDGPU?

[BenPope]

Looking for fury tests, I've also found this result file on openbenchmarking http://openbenchmarking.org/result/1...HA-BPPADOKAN90 that shows a beautiful 221.97 fps on 3440 x 1440 with DRI3. What's the explanation between those differences with Michael's result (on a lower 1920 x 1080), then ? could there be a regression somewhere in between ?

I have no idea why my results are so much better than Michaels. I'll spin a 4.5 DRM-Next in a couple of days and give it another go.

[rabcor]

Not too bad, but... Don't we need this driver to actually surpass catalyst so that it can amount to anything? (Considering how catalyst is lagging pretty far behind Nvidia's drivers)

[magika]

Fury almost got 60 fps in Xonotic.
W O W
O
W

[FireBurn]

I've been really impressed with my Tonga based M395X with the Powerplay branch - I've nearly completed BioShock Infinite running at 4k resolutions

The only things not working for me now is the DPM auto changing of PCIe lanes and runpm - the later of which works without powerplay

I've got bugs open for both and Alex (agd5f) has been helping me debug the issues

The last piece of the puzzle will be the new double buffering for DRI_PRIME which is being worked on for Intel/Nvidia setups and I'm hoping won't take too much effort for Intel/AMD setups

[Adriannho]

The last piece of the puzzle will be the new double buffering for DRI_PRIME which is being worked on for Intel/Nvidia setups and I'm hoping won't take too much effort for Intel/AMD setups

What exactly is that? Does it work on AMD+AMD scenarios?

[Creak]

I hope that during 2016 we'll see the open source AMD drivers surpass Catalyst, so we won't have to compare free and proprietary benchmarks anymore.

[profoundWHALE]

fury almost got 60 fps in xonotic.
W o w
o
w

w o w
o w o
w o w