NVIDIA 387.34 vs. Linux 4.15 + Mesa 17.4-dev Radeon OpenGL/Vulkan Performance

[dimko]

On Nvidia side , it is performing literally same on Windows and on Linux via Wine.

Love to play devils advocate. There are more nvidia owners on Linux. Wine better optimised for Nvidia?

[vein]

Love to play devils advocate. There are more nvidia owners on Linux. Wine better optimised for Nvidia?

Wine has always been optimized for Nvidia...just like basically everything on linux. It is just recently that Amd has come with a viable option... And since it is oss, a lot of people, me included, are very happy for it.

These benchmarks just once again confirm that my 299$ fury card from 1.5 years ago was the best possible buy

[Leopard]

Love to play devils advocate. There are more nvidia owners on Linux. Wine better optimised for Nvidia?

No , because some Amd developer said on Reddit and also here Linux driver lacks some extensions and improvements that Windows driver has.

[OneBitUser]

I think it would be highly beneficial to include at least one previous generation Nvidia card in these tests.
The GTX 970 and 980 were very popular in their day, including either of those would be a good indicator of how well previous generation hardware is holding up.
Plus we could extrapolate the performance of all 900-series GeForces from the performance of a well-known Maxwell card.
The 980 would be my choice, since there was no skullduggery involved with its hardware configuration.

I am not mentioning AMD because their generations are a lot more complicated / muddied.

[Strunkenbold]

There must be some performance regression on Kernel 4.15. My RX 470, paired with a i7 3770k @ 3,9GHz, only 8GB of RAM (DE:MD build ~1GB of cache on the SSD) on the stock kernel 4.13 from Ubuntu 17.10, manages 36 FPS on Deux Ex Ultra benchmark, how can the RX 580 only manage 34 FPS, with a much higher clocked (and 4 more threads) i7 and double (and faster) RAM?

There is already a bug report:

https://bugs.freedesktop.org/show_bug.cgi?id=103100

Ive also did some more tests between old amd-staging kernel and after the bad commit:

https://openbenchmarking.org/result/1712065-AL-GAMETEST356

The most games have only a slight degradation in Performance, Tomb Raider for instance is totally uneffected. But Shadow of Mordor (which has no Benchmark Profile) and Unigine Superposition dropped significant.

Currently I dont know if AMD cares at all. Maybe if more people respond to the bug report it raises the attention..

[M@GOid]

There is already a bug report:

https://bugs.freedesktop.org/show_bug.cgi?id=103100

Ive also did some more tests between old amd-staging kernel and after the bad commit:

https://openbenchmarking.org/result/1712065-AL-GAMETEST356

The most games have only a slight degradation in Performance, Tomb Raider for instance is totally uneffected. But Shadow of Mordor (which has no Benchmark Profile) and Unigine Superposition dropped significant.

Currently I dont know if AMD cares at all. Maybe if more people respond to the bug report it raises the attention..

What do you mean for that? I know there is a benchmark tool inside the game.

But tanks to point me to that bug report. Lets go ask Michael for a quick benchmark on the Kernel 4.15rc3 thread to increase awareness for this bug.

[Djhg2000]

I noticed a few of the benchmarks were only done in 4K. Wouldn't 1080p be more relevant for most readers? Or has everyone else moved up in resulution?

[Michael]

I noticed a few of the benchmarks were only done in 4K. Wouldn't 1080p be more relevant for most readers? Or has everyone else moved up in resulution?

Because in many games at ~1080p with modern graphics cards they just become CPU bound and not really stressing the driver/GPU well.

[Strunkenbold]

What do you mean for that? I know there is a benchmark tool inside the game.

That just means that pts cant make automated results with this game, as there is no profile or in general no support from the game devs for automated benchmark tools.

[pal666]

No it isn't. Try Doom 2016 on Wine and then Windows with Amd. You will see Wine performs worse.

lol, maybe because doom running under wine is different application?

On Nvidia side , it is performing literally same on Windows and on Linux via Wine.

on nvidia side it uses different codepath, the one wine devs optimize against