AMD Gallium3D Marks Huge Win: Beating Catalyst In Steam On Linux Game

[kenjitamura]

The Source engine is using a DirectX -> OpenGL translation layer isn't it? Wonder if that has any weight on the situation. From what I understand a big reason Valve is doing a rewritten Source 2 engine was to include native linux support from the get-go.

[edoantonioco]

This is great news to see how good open source drivers works on TF2, on amd probably is still behind windows, but this benchmark is very interesting. So, keep up the good work, the difference can be higher in a future.

[xeekei]

Why are you such an asshole?
Half life 2 engine games are the eye candy of Linux.
I played Wasteland 2. It doesn't look anywhere near as pretty as games mentioned in the article. Though it is nice game.
I Literally finished HL episode 2 10 minutes ago.
I find this information really relevant.
Just because you don't think those tests are irrelevant, doesn't mean they are.
Many people were asking about HL2 engine games before, and Michael gives those tests now.
Why are you so toxic?

I'd say that Metro: Last Light is the "prettiest" game that Linux has right now. Of course, I'm sure Michael would include it if it provided easy automated benchmarking.

[dimko]

Look, don't mean to be rude, but I don't give a crap about these old ass games.

Why are you such an asshole?
Half life 2 engine games are the eye candy of Linux.
I played Wasteland 2. It doesn't look anywhere near as pretty as games mentioned in the article. Though it is nice game.
I Literally finished HL episode 2 10 minutes ago.
I find this information really relevant.
Just because you don't think those tests are irrelevant, doesn't mean they are.
Many people were asking about HL2 engine games before, and Michael gives those tests now.
Why are you so toxic?

[liam]

Yup, 3 posts. So what? Bite me.

They aren't though. Benchmarking two games is hardly indicitive of the Linux gaming scene. I can't be the only one interested in ditching Windows (finally) for gaming, and am curious how it performs without having to do a wipe and see for myself (which I've done numerous times).

According to a reply I received from Aspyr, they're blaming a lot of the Borderlands 2 performance issues with the AMD driver, so I was hopeful that maybe Gallium would be better (will be better?)

You don't need to do a wipe. You've got several options, from wubi(assuming that's still around), side-by-side install, virtualization, or good ole live USB.

[artivision]

You would lose that bet.

My Radeon-6870 with Gallium and Nine smokes Catalyst everywhere from Linux to Wine gaming.

[Mikeyy00]

You should try a little harder.



Just joined, only 3 posts and you aren't talking with very much respect. These are as up to date as you can get, so I don't even know why you're complaining.

Yup, 3 posts. So what? Bite me.

They aren't though. Benchmarking two games is hardly indicitive of the Linux gaming scene. I can't be the only one interested in ditching Windows (finally) for gaming, and am curious how it performs without having to do a wipe and see for myself (which I've done numerous times).

According to a reply I received from Aspyr, they're blaming a lot of the Borderlands 2 performance issues with the AMD driver, so I was hopeful that maybe Gallium would be better (will be better?)

[Mikeyy00]

CS:GO is new game for Linux, appeared just in less then 2 months. Wasteland i little older for that matter, Borderlands 2 has similar date and Civ 5 is much older - they are all older really isn't it

Ah you mean engine is probably older , but yeah i agree it should be probably the best to test different games, but one choosed game maded upon different engines - so one game per engine, that should better show how driver really performs across the board

I think that way (when most engines are covered) most people will stop to complain how something is "old"

Sorry, wasn't trying to be a dick earlier. It just seems that the Valve games are the only ones mentioned. I wasn't aware there isn't a way to benchmark the other games.

That being said, I'm interested to see if they are playable on Gallium, since Catalyst is a mess (compared to NVidia).

[dungeon]

Look, don't mean to be rude, but I don't give a crap about these old ass games. They've been running decently on Gallium for a while (as long as you're not rocking an ancient adaptor). What about newer games like Civ 5? Borderlands? (for which the Catalyst runs extremely poorly?). How about Wasteland 2?

CS:GO is new game for Linux, appeared just in less then 2 months. Wasteland i little older for that matter, Borderlands 2 has similar date and Civ 5 is much older - they are all older really isn't it

Ah you mean engine is probably older , but yeah i agree it should be probably the best to test different games, but one choosed game maded upon different engines - so one game per engine, that should better show how driver really performs across the board

I think that way (when most engines are covered) most people will stop to complain how something is "old"

[zanny]

Don't know why people keep asking for Witcher or the Aspyr games. PCGamingWiki has already shown they do not have anything close to performance parity with Windows on Nvidia, and if that is not the case it just usually means a poorly optimized engine.

Unless Civ 5 / Borderlands got engine optmizations (and they will probably will) to give us even a hint that the engines are anything close to well performing, then benchmarks on them might be worth something. And Witcher 2 will never not be DirectX wrapped with a proprietary translation layer, so I can't see how those benchmarks would be useful at all.

The reason is that you cannot really hold drivers accountable for terribly written OpenGL. If engines are poorly implemented, they might have tremendous performance disparity between drivers, but that might just be because, say, Catalyst has internal logic to mitigate the poor implementation, designed specifically for that game, but we can't know that.

For better or worse, Xonotic remains probably the most optimized OpenGL game on Linux (at least from source review, Unigine might be good too) and that means it produces the best "real world results", assuming you can tax the GPU enough running it to keep it at framerates that matter. If you are benchmarking it at 300+ fps it does not really mean anything.