For anyone interested in some gaming benchmarks, Gamer's Nexus (a reputable source with good methodology) has some numbers here: https://www.youtube.com/watch?v=ovOx4_8ajZ8

Based on their results, it sounds like there's still quite a way to go Linux gaming/Proton (ie: very inconsistent frametimes on Nvidia hardware), but it's definitely been taking steps in the right direction.

▲

kachapopopow an hour ago | parent | next [-]

I actually get higher FPS in graphically demanding games due to vulkan producing more optimized shaders, most of the lag actually comes from context switching and translation for directx, it's absolutely GREAT one dx12 due to how little translation there is.

I'm currently working on few very targeted optimizations for several hotspots I've found out from messing around so it will be interesting if I can solve those horrible stutter issues on cyberpunk since if I can fix it (in a ghetto and janky way), so can valve.

I can also achieve higher fps in games like overwatch (dx12) out of the box on nvidia on proton experimental which is surprising as 4 years ago the input latency was unbearable and I had drops as low as 30 fps, now I can achieve consistent 600 fps with minimums of 450 whereas on windows I get as low as 220fps and averages of 500ish. I do have anticheat related drops to <300fps due to the amount of translation happens when they decide to scan memory, registry and whatnot although it lasts <1s and doesn't happen during games it seems.

▲

d3Xt3r 3 hours ago | parent | prev | next [-]

That's mainly an nVidia issue. On AMD, Linux is actually faster and more stable compared to Windows.

▲

Night_Thastus 3 hours ago | parent [-]

There seems to be little evidence of that, at least from a reputable source.

For some games it can be. For some games Proton performs far worse than Windows. It's not steady across the board. And some have stability issues, bugs, major performance problems, or just flat out don't work.

I want Proton to be the future as well, but I think it's important not to oversell it as a drop-in replacement either.

EDIT: GN highly recommends against apples-to-oranges comparisons of the two, but even looking at their own data for AMD cards (with exact same CPU, RAM, and motherboard) it clearly shows Proton being behind on the order of 6-15%. Not a lot, but not ahead either. You can compare the numbers for the AMD cards against this video's here: https://www.youtube.com/watch?v=yP0axVHdP-U

EDIT 2: Instead of just down-voting because the result makes you unhappy, how about responding with well-sourced proof otherwise?

▲

birksherty 2 hours ago | parent | next [-]

Another video showing the difference: https://www.youtube.com/watch?v=fqIjUddUSo0

▲

dralley 3 hours ago | parent | prev | next [-]

From personal experience, all of the games that I personally play seem to work at least as well if not better on Linux than Windows. The only exception is FSR4, but that ought to be fixed soon.

▲

d3Xt3r 2 hours ago | parent | prev | next [-]

FWIW, I wasn't the one who downvoted you. I can't even downvote here as I'm apparently "karmically broke". I have 468 karma, not sure how much I even need to downvote...

▲

simoncion 2 hours ago | parent | prev [-]

> There seems to be little evidence of that, at least from a reputable source.

I dunno. I remember a little while back some reviewers got a hold of both a Windows version and Linux version of a handheld gaming machine that had exactly the same hardware. The conclusion reached was that the Linux version was better in nearly every way.

As I remember it, a little while after this happened, some muckety-muck in the Gaming Division in Microsoft announced something like "We're making a new committment to consistent, high performance in Windows on handheld gaming devices! We're going to ensure all those little game-spoiling roadblocks are removed!". Which, like, good job making it NOT look like you're spasmodically reacting to bad press, guy.

▲

Night_Thastus 2 hours ago | parent [-]

Handhelds and desktop PCs can perform differently. It's very possible for Proton to perform better for handhelds but worse for desktop PCs.

	▲	simoncion 2 hours ago \| parent [-]
		> Handhelds and desktop PCs can perform differently. The handhelds we're talking about are -essentially- a low-power [0] laptop in a tiny case. Again, we're talking about exactly the same hardware, that provides substantially worse performance when Windows is the OS than when Linux is the OS. For reference, here's one instance of the original coverage of the phenomenon about which I spoke: <https://www.youtube.com/watch?v=CJXp3UYj50Q> [0] But higher power than one might naively expect!

▲

simoncion 2 hours ago | parent | prev [-]

> Based on their results...

Do note that they make it very, very, very clear that their results are preliminary, and while they've put a whole lot of work into setting up benchmarking on Linux, they're not at all sure that they've got it all correct.

> ie: very inconsistent frametimes on Nvidia hardware

Yeah, Nvidia on Linux for non-"compute" use has always been a terrible, godawful shitshow. Given Nvidia's recent and fairly-clear disinterest in the "selling graphics cards for people to play video games with" market, [0] I can't imagine things will get consistently better anytime in the near future. [1]

[0] Why sell that silicon to video gamers when you can sell it to cryptominers and -these days- "AI" companies for a much, much, much higher profit?

[1] I mean, it took them how many Windows driver revisions for them to release somewhat-non-garbage drivers for their spanking-new fancy-ass 5000-series cards? And Windows is the only consumer OS that they care about! For video game use, the Linux drivers are gonna be starved for development resources, and -unlike ATi/AMD's drivers- noone in the world can work on them but Nvidia.