Radeon RADV vs. NVIDIA Vulkan Performance For F1 2017 On Linux

[pal666]

Nvidia has offered basically the same Windows performance to Linux users and what do they get in return? "You suck, because your driver is closed." Doesn't that sound at least a little bit awkward?

that sounds deserved, unless you are nvidiot
nvidia didn't offer same performance - linux games are usually slower. so if you value only speed, you are not using linux in the first place

[jrch2k8]

Well to be honest AMD right now have a driver that is perfect for 95% of gamers in linux(Freesync/regular 1080p60+ monitors) that won't care a loss of few FPS here or there and value open source and next gen features like be able to use Wayland compared to nVidia and the same can be said for Windows at many price brackets(not all, specially with the crazy mining prices).

nVidia is still the best for the ANAL linux gamer(and windows as well on the top segment), aka those that either get off watching FPSs instead of gaming or those that genuinely have more specialized hardware(Gsync/Regular 144Hz+ monitors for example) where those few FPS lost can make an actual difference at gameplay quality.

[lucrus]

Nvidia has offered basically the same Windows performance to Linux users and what do they get in return? "You suck, because your driver is closed." Doesn't that sound at least a little bit awkward?

You tell it as if NVidia were a charity. Nvidia hasn't "offered" anything to Linux users. It has sold some products for cold, hard cash. And that's exactly what they get in return: cold, hard cash, but that's all.

What about love? They don't deserve it. NVidia releases signed blobs (let alone free software or documentation) only when they are forced to. They are forced to release their shit when they need a patch accepted in Linux and the only way is releasing the signed blob, so that nouveau developers can provide a FOSS use case for their f**king patch. In other words, they release it for themselves, not for me, not for you and not for Linux users in general.

AMD drivers are slower? Price/performance is worse for AMD cards? Who cares? AMD cards are good enough (more than enough in my case), and I'm more than happy to vote for them with my wallet. And not only my wallet: my friends ask me from time to time what they should buy. I don't even take NVidia hardware into account for them too.

Of course everyone is entitled to look only at price/performance ratio and to keep allowing NVidia make a fool of him/her; go ahead. I prefer waiting by the river.

[sarmad]

nVidia is still faster, but the gap is steadily narrowing. At this point I think AMD is unquestionably the recommended choice for Linux gamers.

[bridgman]

Just an FYI, I will have more low-end cards (and more card tests in general), tomorrow or Sunday. Simply ran out of time today before deciding to cut it and run an article to avoid something not too late on a Friday to publish, but yeah larger comparison over the weekend among other tests.

Excellent... figured it was something like that. You usually have a good reason

[sa666666]

Nvidia has offered basically the same Windows performance to Linux users and what do they get in return? "You suck, because your driver is closed." Doesn't that sound at least a little bit awkward?

While the presentation could be a little better, I think this reflects the preferences of people using Linux and open source. Of course they will be against closed source and advocate for open source whenever possible. If not, they probably wouldn't be using Linux.

I personally used Nvidia for years in Linux, because there was nothing better from anyone else. But always in the back of my mind I couldn't wait to move to an all-open source system. Now that AMD makes that possible, I've moved fully to AMD (CPU and GPU).

It's kind of specious to expect open source advocates to be thankful for closed source, proprietary drivers. We want everything to be open; that's why we're using Linux in the first place!

[Michael]

4. Michael included a few lower-end AMD cards (550, 560, 285) but no corresponding NVidia cards, so AMD appears generally lower overall than NVidia

Just an FYI, I will have more low-end cards (and more card tests in general), tomorrow or Sunday. Simply ran out of time today before deciding to cut it and run an article to avoid something not too late on a Friday to publish, but yeah larger comparison over the weekend among other tests.

[bridgman]

Still, you tell me if those AA graph look like Nvidia is doing only "marginally better".

Thanks for suggesting that - the first thing I noticed is that there are not any graphs for "full workload" (4K, AA on) which is where I would expect our hardware to do best.

What I see is the following:

1. The NVidia driver hits what appear to be CPU limits at a higher FPS point than radv, so with light GPU workloads all the NVidia cards are faster than all the AMD cards

2. As the workload increases the AMD cards move up and start to interleave with the NVidia cards

3. We ran out of graphs before getting to full workload (4K, ultra-high or AA on or both). I'm not sure that the pattern of "moving up with increasing workload" would continue but based on results from other games it seems pretty likely

4. Michael included a few lower-end AMD cards (550, 560, 285) but no corresponding NVidia cards, so AMD appears generally lower overall than NVidia

BTW something I should add in your favour... there is nothing inherent about DX11-over-Vulkan that makes it *impossible* to write a game which would take full advantage of AMD HW; it's just that typical performance tuning work on DX11 will not lead you there.

[bug77]

Just curious, why do you think the 580 should be running circles around the 1060 on this game, particularly at resolutions where FPS is sufficiently high that per-frame overhead starts to dominate over actual graphics work ?

You can't use the flops argument here since AFAICS the game was not written around one of the modern APIs (DX11-over-Vulkan doesn't count) and so will probably have not been written to make full use of the compute throughput of the 580.

I was thinking how AMD's hardware tends to perform better under DX12 and Vulkan because of the lower overhead of async tasks (iirc). But you're right, this isn't a pure Vulkan title (as I, myself, posted several times here), so the performance is about where it should be.
Still, you tell me if those AA graph look like Nvidia is doing only "marginally better".

[bug77]

yeah, we must not praise AMD until it is clearly better than NVIDIA, we cannot just praise it when the games are playable and have pretty decent framerates to be enjoyable.

Praise them all you want, I'm not saying otherwise. I give them credit for what they have accomplished. But performance? They're only starting to get where they should be.
Nvidia has offered basically the same Windows performance to Linux users and what do they get in return? "You suck, because your driver is closed." Doesn't that sound at least a little bit awkward?