This Chart Shows How The Radeon RX 580 vs. GeForce GTX 1060 Now Compete Under Linux

[humbug]

Polaris drivers seem to be in great shape, well done to everybody who worked on it!

Hope that the R9 290 RadV performance gets optimized more. Seemed to under perform in this game based on Michael's numbers.

[zanny]

If the stars align and Navi isn't instantly out of stock due to miners next year it might finally be time to recommend Linux + AMD to new system builders, since dxvk is advancing so fast by then it should be possible to run Windows games at near-equal performance. What a time to be alive.

[Xicronic]

I know it's easy to see AMD's average beating out Nvidia and celebrate, but...
Looking at high: min / avg / max:
1060: 55 / 76 / 105
580: 37 / 94 / 149

The 1060 has a phenomenal lead in minimum fps. For me, never dropping below 55 would be far preferable to dropping to 37 but having higher average fps. I think most people would agree it also feels better to play a game when fps does not jump between a high average and a low min.
Maybe it was a one frame fluke, but we don't have any 99th percentile frame data. So, given the information presented to me, I still think Nvidia would be the better choice in this situation.

[monte84]

As far as the minimums go, I think that is just initial load in. I have not seen those types.of dips in actual game play. So looking at the minimums alone can be misleading.

[chuckula]

Given the raw theoretical hardware capabilities of the Rx 580 are practically equal to the GTX-1070 and the power consumption numbers are often worse than the GTX-1080, there still needs to be some work done.

[valici]

I know it's easy to see AMD's average beating out Nvidia and celebrate, but...
Looking at high: min / avg / max:
1060: 55 / 76 / 105
580: 37 / 94 / 149

The 1060 has a phenomenal lead in minimum fps. For me, never dropping below 55 would be far preferable to dropping to 37 but having higher average fps. I think most people would agree it also feels better to play a game when fps does not jump between a high average and a low min.
Maybe it was a one frame fluke, but we don't have any 99th percentile frame data. So, given the information presented to me, I still think Nvidia would be the better choice in this situation.

When grasping for straws...

In windows in this title, the 1060 is much faster.. in Linux the 580 is much faster. It's pretty clear that AMD is the way to go.

[artivision]

Something is very wrong with Radeons above 6-Teraflops, even on Windowz they don't perform as well as other Radeons under 6-Teraflops Flop/Flop comparison, wile the over 6 ones use only HBM...

Something is very wrong with Geforces to, the "good" ported games and the "well supported" DXVK ones always give 85% versus native Windowz, is this a magic number or what?

[Leopard]

But it wasn't the case on Windows already?

580 roughly sits between 1060 and 1070

[dwagner]

Polaris drivers seem to be in great shape

Regarding 3D performance, that may be true - but in terms of stability they are absolutely not ready for production use. I would so much more like to see amdgpu win an uptime contest rather than some game benchmark...

[juno]

Given the raw theoretical hardware capabilities of the Rx 580 are practically equal to the GTX-1070 and the power consumption numbers are often worse than the GTX-1080, there still needs to be some work done.

Look again, the 1070 has more raw power compared to a 580.