Radeon Windows 10 vs. Linux RadeonSI/RADV Gaming Performance

[liam]

Unfortunately AMDGPU-PRO doesn't work on 16.10 yet and was already a ton of work doing this comparison as is.... (and judging from the NVIDIA article, I only got 2 or 3 new subscribers, so really barely covered costs for all of that benchmarking as is and remains to be seen if will get any new subscribers or donations from this article)

You've said this before.
Is your reasoning that you expect new subscribers to pay for new articles? If so, that suggests no consideration is given to readers once they've subscribed.

[humbug]

That is the ideal situation that I think alot of us gamers are hoping for. I have the suspicion in the know devs already intuit what AMD's plans for open sourcing their Vulkan code is. I have to assume that Radv's very existence is evidence of that. Public understanding of it has been slower to get out then I would have liked though. A timeline would be cool. I assume porting companies already know through their contacts at AMD what the Vulkan plans for linux are too.

It's just a matter of time, but how much time I think is still not public knowledge yet.

so you are saying that because there is no still no performant open source vulkan driver for AMD that game developers cannot target vulkan? I guess AMDGPU pro is not a viable alternative as the distro and kernel support is so limited, although it is way faster than RadV.

If you are right and AMD's open sourcing of vulkan driver is not gonna happen anytime soon then maybe that's why Valve is so eager to improve RadV.

[bridgman]

You've said this before. Is your reasoning that you expect new subscribers to pay for new articles? If so, that suggests no consideration is given to readers once they've subscribed.

I think it's the other way round... writing articles aimed at existing subscribers but hoping they will attract new subscribers... all the while keeping an eye on what actually happens.

[suberimakuri]

If you are right and AMD's open sourcing of vulkan driver is not gonna happen anytime soon then maybe that's why Valve is so eager to improve RadV.

I don't know anything nor anybody in this graphics world, but I suspect this.
I think AMD's opening of OpenCL to be interesting and somewhat backs up the current RADV situation. ROCm to me came out of left field but in any case I'm grateful for it.

[rabcor]

Damn, radeonsi surprises.

[Michael]

You've said this before.
Is your reasoning that you expect new subscribers to pay for new articles? If so, that suggests no consideration is given to readers once they've subscribed.

No, that's not the case. I routinely run tests quite frequently solely at the request of Phoronix Premium readers, often times where there is just a sole premium member contacting me with said request. Just with the Windows articles are sort of a special case because 1. I need to do all the Windows setup since I don't keep Windows installs around 2. lots of manual testing involved... Like these two articles basically took a week of work and aren't really sustainable in their own right. With the usual premium reader requests where it's with usual PTS testing, etc, it's rather straightforward to carry out, happy to do so, etc.

[Solid State Brain]

- There is no need to do these tests very often; once every 1 or 2 months would be more than enough given that driver improvements do not come that fast either.
- Perhaps for these AMD OS comparison tests in order to reduce the work you could limit the amount of games to a handful popular ones with the most commonly used settings (e.g. 1920x1080 high detail) that are known to perform worse in Linux. This could cut your work to less than half.
- I can't believe you don't have a spare hard drive or even just system image to keep a complete Windows system installation on!

PS: I'm not a subscriber, but I liked the article and I rewarded it in other ways (*wink*).

[Michael]

- I can't believe you don't have a spare hard drive or even just system image to keep a complete Windows system installation on!

It's about always using the same hardware, even the same hard drive, throughout all of the tests. And so no, I don't have an extra 950 PRO NVMe SSD... I wish I did. And last I knew, Windows wasn't so friendly when swapping out an installation between systems and booting it up on a completely different system, has anything changed with Windows since?

[DanT]

When comparing the results (not reading the text at the same time), it is often difficult to figure out which Windows driver is being used (OGL, DX11 or DX12). In some cases, for example Tomb Raider and Shadow of Mordor, I didn't find that info at all.

Yes, I often have the same difficulties.

Just a low-priority feature enhancements request: It would be nice, if PTS could always add OpenGL/Vulkan/DX labels to the graph legend (this information should be in configuration of the test anyway, shouldn't it?). I believe that it will increase the readability of the results very much.

[Michael]

Yes, I often have the same difficulties.

Just a low-priority feature enhancements request: It would be nice, if PTS could always add OpenGL/Vulkan/DX labels to the graph legend (this information should be in configuration of the test anyway, shouldn't it?). I believe that it will increase the readability of the results very much.

PTS only ran a few of the tests in the article since all the others had to be done manually due to limitations of the games.