Wine-Staging Has Been Revived, Working Towards New Release

[iyxwsoekthsv]

You don't know what you're saying

Actually I do. Thing is, what I am saying is from actual real world experience. Not talking about coding theory, about context switching and what either should perform like. Talking about what I'm actually experiencing in terms of FPS, frame times, input latency, real world performance vs theoretical performance.

The primary difference I've always experienced is Gallium Wine being far more jittery. Sure, it may peak as well or even better than CSMT but it's far more jittery. CSMT for me personally at least has always been the more smooth experience. Can't tell you exactly why that is. Just telling you what I've experienced.

There could be tonnes of different factors contributing to that; maybe it's the fact I'm not on the stock CPU scheduler (CFS). Maybe it's as a result of using pure AMD hardware. Maybe it's something else entirely. Just telling you, I personally have not felt any improvement from using Gallium Wine. I'll grant you the compatibility though. There have been titles that only ran in Gallium Wine. Then again, there have been far more titles that ran in not a single version of Wine. Stable, Devel, Staging, Gallium or otherwise.

[Naquatis]

Actually I do. Thing is, what I am saying is from actual real world experience.

I have a nice example that shows that Gallium Nine works much better than CSMT. I am using Gallium Nine on RX 480 and playing the 64 bit version of Tera (it is using Unreal 3 engine). My real world experience tells you it is like day and night! With CSMT you did not like to go into any instance because Wine begins to stutter while loading textures and has a much lower framerate. With Wine + Gallium Nine I never got this problems!

Another example is Guild Wars 2 where you really like to have Gallium Nine. It is the same day and night feeling.

[iyxwsoekthsv]

My real world experience tells you it is like day and night!

Not discounting your real world experiences. Just don't discount mine either. Not going to list examples, there are too many.

But, it's good either way that for you Gallium Wine does do what it's supposed to do. The more people that can use Wine to move away from Windows, the better.

[mannerov]

EDIT: The problem with gallium nine was the original authors refused to budge on their ideology so they couldn't get the wine bits upstreamed. And upstream has real valid concerns about it. The fact that it couldn't get upstreamed and the authors have ideological issues prevented real flaws from getting fixed. But regardless of all that it is still the better option, even today.

Please don't say such comments on topics you don't know actually what happened. Reality is completly different story than yours. Thanks.

[leipero]

Talking about what I'm actually experiencing in terms of FPS, frame times, input latency, real world performance vs theoretical performance.

The primary difference I've always experienced is Gallium Wine being far more jittery. Sure, it may peak as well or even better than CSMT but it's far more jittery. CSMT for me personally at least has always been the more smooth experience. Can't tell you exactly why that is. Just telling you what I've experienced.

There could be tonnes of different factors contributing to that; maybe it's the fact I'm not on the stock CPU scheduler (CFS). Maybe it's as a result of using pure AMD hardware. Maybe it's something else entirely. Just telling you, I personally have not felt any improvement from using Gallium Wine. I'll grant you the compatibility though. There have been titles that only ran in Gallium Wine. Then again, there have been far more titles that ran in not a single version of Wine. Stable, Devel, Staging, Gallium or otherwise.

My experience is total oposite of yours, and I'm not sure how much impact hardware does have aside from GPU, for example, one game, while I was using GT210 (with nouveau) on another machine wine CSMT would result in much better performance compared to gallium nine, however when some "billion blocks" maps are loaded it would result in stuterry mess on CSMT while nine did worked terrible anyway..., so no matter what, you end up with bad results, not because GPU is limited (very same GPU works fine under Windows ofc., no stutters), but because you get CPU bottleneck, and it doesn't matter even if you use some very high end CPU when game is single threaded (unless all translation is moved to other threads, and I don't think that's the case). With Radeon (even r600) it's different story, with wine CSMT you get better performance compared to nouveau simply because r600 is better driver, but it's still in 30-40 FPS range on "billion blocks" maps, while when gallium nine is enabled you get into 60-100+ FPS range, that's like double, with lower resolutions (720p) wine CSMT doesn't move up as much, it is still in 40 FPS range, while gallium nine goes 90+FPS.

Sometimes, when wine gets some bugs in d3d (in 2.x versions was the case) and some game would refuse to work, it would actually work with gallium nine, it happened to me, can't remmember what was game. Just tested GRID this morining, and I did not measured FPS, but with wine 3.2 (CSMT included) ultra settings 4xAA, 1080p it was probably in 30FPS range, playable, but with some "pauses" when loading menu etc., gallium nine 60+ FPS ultra smooth.

There's only one DirectX game that I did found it works better with wine CSMT compared to nine, and that's Split/Second, but only on high/v-high settings, on low settings nine is better even there, and the reason why it doesn't work well with wine is aincent hardware where some things arrent implemented in r600.

[iyxwsoekthsv]

My experience is total oposite of yours, and I'm not sure how much impact hardware does have aside from GPU

Well, hardware obviously has a big impact. My hardware is shit either way. It's an AMD APU, no discrete GPU. So, either way I'm hardware capped, hard. It's been a while since I did actual measurements though. It may be interesting to revisit Gallium Wine, just to see how it compares to, say, 18 months ago or so. That's basically the last time I did consider it seriously for gaming purposes.

Developments in the Mesa stack and developments in Wine in general may have seen some relevant improvements that would make Gallium Wine perform as well for me as it apparently does for others. Will try it. Bored anyway, so, why not, ey? Would be nice to eke out ever so slightly more out of this aging hardware.

[Naquatis]

Tomb Raider 2013 runs faster with CSMT than Nine and you can use the DX11 mode that renders the game better. You can not run the AC 4 Black Flag DX10/DX11 game and other DX10/DX11 games with nine. Nine does not work with nvida and intel gpus.

I am ok with that! I did not say I like to replace one with another and I am also not the guy who say do everything to remove Windows. I like diversity and before I try to use Wine for gaming I would use Steams inhome streaming. ;-)

[leipero]

Well, hardware obviously has a big impact. My hardware is shit either way. It's an AMD APU, no discrete GPU. So, either way I'm hardware capped, hard. It's been a while since I did actual measurements though. It may be interesting to revisit Gallium Wine, just to see how it compares to, say, 18 months ago or so. That's basically the last time I did consider it seriously for gaming purposes.

Developments in the Mesa stack and developments in Wine in general may have seen some relevant improvements that would make Gallium Wine perform as well for me as it apparently does for others. Will try it. Bored anyway, so, why not, ey? Would be nice to eke out ever so slightly more out of this aging hardware.

Well, my hardware is sh*t also, FX CPU, that's where nine helps the most, 18 months ago, let me tell you something, when I've first tried gallium nine, it was in fglrx days, and I did prefer CSMT over nine because nine was buggy as hell and not worth it, but improvements made it different. You should definitively try it again, assuming you did everything properly you will be suprized (but as I noted, it's not all perfect ofc., one example Split/Second, but there are probably more examples).

[duby229]

Actually I do. Thing is, what I am saying is from actual real world experience. Not talking about coding theory, about context switching and what either should perform like. Talking about what I'm actually experiencing in terms of FPS, frame times, input latency, real world performance vs theoretical performance.

The primary difference I've always experienced is Gallium Wine being far more jittery. Sure, it may peak as well or even better than CSMT but it's far more jittery. CSMT for me personally at least has always been the more smooth experience. Can't tell you exactly why that is. Just telling you what I've experienced.

There could be tonnes of different factors contributing to that; maybe it's the fact I'm not on the stock CPU scheduler (CFS). Maybe it's as a result of using pure AMD hardware. Maybe it's something else entirely. Just telling you, I personally have not felt any improvement from using Gallium Wine. I'll grant you the compatibility though. There have been titles that only ran in Gallium Wine. Then again, there have been far more titles that ran in not a single version of Wine. Stable, Devel, Staging, Gallium or otherwise.

I didn't mean to imply you don't have your own subjective experience. But the real fact of the matter is that benchmarking already proved gallium nine has much better CPU performance characteristics, by hundreds of percent. Multiple times better. Jitter may well be 99th percentile frame times, and perhaps Wine CSMT is able to achieve better metrics in that scenario, I don't know. But I can say I haven't noticed that for myself, generally Gallium nine gives much better minimum FPS and that's what people notice.

EDIT: The graphics stack has many layers and wine CSMT is one of them. And it loads the CPU -hard-, super hard. Because it removes so many CPU cycles, and loses many times more to context switching, it bottlenecks all the other parts of the graphics stack.

[duby229]

Please don't say such comments on topics you don't know actually what happened. Reality is completly different story than yours. Thanks.

Please share what you know. We'd all be interested in hearing it.