- Joined
- Dec 17, 2019
Fixed that for ya.Anyone who has a general-purpose PC expecting their 8GB cards to game will be sorely disappointed.
See, I hate """benchmarks""" like these as they are cherry picking games and then putting a sensational headline that doesn't even line up with their cherry picked charts.we're seeing SteamOS outperforming Windows at running ProtonDB platinum-rated Windows games.
-Returnal is an outlier that runs way better on the 'eckOS than it does on Windows. That doesn't mean it's pulling performance out of it's ass, it means that this specific game has some odd performance issue on Win11.
-Win11 on Asus drivers actually outperforms Linux in Borderlands 3 by a margin. But no one would use that to say that Linux is completely dogshit for gaming since "that's an outlier dude". Rules for me not for thee.
-Homeworld 3 has a very small improvement on Linux, with the Lenovo drivers clearly struggling compared to Asus ones, where the Asus one matches Linux within the margin of error in the lower presets.
Cyberpunk 2077 and DOOM are definitely games where Linux's way of dealing with the games gives it an advantage, but again, an outlier that means Linux does something better in these specific games than Windows. Maybe DXVK is doing some heavy lifting to make it run better for example. Most importantly: all of these tests are done on handhelds that can run SteamOS where all the performance is arguably dogshit on both if we're talking PC gaming, since that's where people want Linux. On full ATX desktops.
This is why I have a problem with all of these Windows vs Linux benchmarks. They're not objective, they're not informational. They'll cherry pick a few select games and twist the results in such a way that they can claim Total Linux Victory and it's so disingenuous that it's just adding fuel to the dumpster fire that is "Linux community's reputation in the common consensus".
What I'd like to see, is a massive test of all kinds of games. New, old, DirectX 9 to DirectX 12. Vulkan and OpenGL. All of them done on a decent PC rig, something with a fully sized RTX 3060 and an AMD equivalent to compare how both GPU vendors are handling themselves under both OS'. A massive dataset where you can't do some smarmy journalistic data manipulation, just an objective megachart that'll show how every game compares. And most importantly: do the best possible job at figuring out why a given game runs better/worse on Linux/Windows. None of this "unlocking hardware potential with the superior OS" bullshit that all these articles peddle. The only reason a game runs significantly worse on one of the two operating systems is some sort of bottleneck. Is it the API where DXVK manages to circumvent whatever's choking it under Windows? If so, can you replicate it under Windows by dropping in DXVK as it is a Windows native binary that also works under Windows? And if it's better/worse by a margin, is it an actual improvement or just a margin of error? Re-run those tests a few times to see if it's just small fluctuations or constant improvement patterns.
Unlike what most people ITT think, no, I don't hate Linux gaming. I'm actually pretty happy that there's some sort of alternative forming in this space, but I'm more interested in the technical aspects of it, since this can be a mutually beneficial race where both platforms reach the same base performance level in all games. It's clear that some things Windows does worse, and some things Linux does worse. Now it's to figure out what both are doing worse so that both can be better. That's what I'd like to see from these comparisons but that's never what I get, I only get half-truths, manipulations, lies and emotionally charged articles worthy of mainstream