Steam Linux Usage Reportedly Ticks Up To 0.8% For August

[phoronix]

Phoronix: Steam Linux Usage Reportedly Ticks Up To 0.8% For August

Due to the US Labor Day holiday, Valve was slow in updating their monthly figures for their controversial Steam Survey of hardware/software data by polled users. At least for their initial batch of August numbers they are reporting a small increase in the Linux gaming population...

http://www.phoronix.com/scan.php?pag...or-August-2019

[ThoreauHD]

That's a big upswing out of nowhere. Kinda wonder if the metrics are BS in the first place.

[arQon]

The problem with WINE / Proton / etc, which too many people constantly refuse to acknowledge, is that it's simply not good enough to have even MOST of your games work in it. People being able to switch to it is an all or nothing deal, and when WINE's track record for AAA titles with sales in the tens of millions is as poor as it is, that's just not going to happen.

To take an easy example: anything on Bethesda's current engine. (Blame the game developers all you want if that makes you feel better about it, but it doesn't change the outcome). Fallout 4 sold *12 million* copies on day one alone... in 2015. And it's still rated *Garbage* on WINE, 4 years later. So it's great for you if all YOUR games work in WINE, or you only play Linux-native titles. But saying things like "90% of the top 100 games work in Proton now, so Linux gaming is finally here!" is just hopelessly false, regardless of how much we would LIKE it to be true.
You can't just look at the math naively (or wilfully deceptively), because what that 90% (not the real number BTW, just convenient to illustrate the point) really means is that even someone with only a handful of games very rapidly reaches a point where they only have a single-digit percentage chance of being able to use WINE without having to dual-boot for at least one game - and as long as that number is non-zero, switching just isn't going to happen.

It doesn't help that the WINE devs' response to someone producing a DX compatibility layer vastly superior to theirs, by himself and in literally a tenth of the time WINE has had, was basically "He's doing it wrong, we have our own plan and ours will be better". History rather strongly contradicts that stance, and even if it is then given their track record most of us will be dead by the time they ever get there. :P

I'd LOVE to be able to drop Windows (especially with 7 now just months from EOL, and Spyware10 about to become the only option) but for all that even this current "still less than 1%" is genuinely a significant uptick proportionally, doing so just isn't viable for me the same as it isn't for more than a tiny fraction of gamers, even if they're aware of Linux in the first place. That "Anything less than ALL my games is no different to NONE of them" problem is why it's not going to get better than 1% for a very long time, even with the resources that Valve and the community are putting into it and the fact that the situation has actually improved greatly in the last year.

[Almindor]

The problem with WINE / Proton / etc, which too many people constantly refuse to acknowledge, is that it's simply not good enough to have even MOST of your games work in it. People being able to switch to it is an all or nothing deal, and when WINE's track record for AAA titles with sales in the tens of millions is as poor as it is, that's just not going to happen.

<snip>

Thing is it's still just an emulation layer with all the downsides. It solves the terribad linux "shared object of hell" and other protocol incompatibilities (I'm looking at you invalid dbus message), but it breaks on its own. See for example No Man's Sky and it's latest update which broke the game completely, even tho it worked 100% before.

Why? Coz nobody tests for it of course. It's a dead end.

[Almindor]

I feel like they should make this value into a source of entropy for some RNG. Those who break it shall be our prophets of the 1%!

[LeJimster]

The problem with....

I get where you're coming from, but Proton for the most part is a step ahead of vanilla wine with support. As more work is being done less and less workarounds are being needed for individual titles. You mention Fallout 4 - https://www.protondb.com/app/377160 - seems like some extra tweaks are required but it can be run with good performance on Linux. Some titles may need a lot of extra tweaks/workarounds and some are just impossible right now because of Anticheat software. It may never get to 100%, but I do believe we will see over 90% support in the coming years. Right now we are sitting at around 60% of titles working with minimal to no tweaking, which is very impressive considering Proton is a little over a year old from it's first release.

I don't have a massive steam library, but well over 90% of my games are playable on Linux right now and some that were unplayable stuttery mess (even though they ran) now play flawlessly thanks to dxvk improvements and things like Valves ACO compiler. So the improvements keep on coming, but Linux gaming is still an adventure and it's not everybodies cup of tea.

[Templar82]

Who says that? As i see, Top 100 is @ 60%, by somewhat vague people votes but OK that

76% of top 1000 is rated Silver or higher which 'should' mean "Runs with minor issues, but generally is playable" or better


Is there a site that tracks which games don't work on Windows 10? Because there are at least a few of those.

[Xaero_Vincent]

The biggest headache for Wine is the mass adoption of anti-cheat software in multiplayer games. Other than that, I think it has a higher success rate than a miss rate these days.

[arQon]

@Jimster - thanks for digging out the correct percentage. (and @ almost everyone else: I said "NOT THE REAL NUMBER"! Come on, guys...)

A 60% "working" rate for the Top 100 means that if you play a mere FIVE games in that list, your chances of actually being able to play all of them are less than 8%. That's what I find so frustrating about the cheerleading every time there's a Proton article. Don't get me wrong: it's great to be supportive of the efforts, and fine to be optimistic about them, but neither of those things is an excuse to pretend that there isn't still a very steep hill to climb.

And that "8%" outcome ignores how dishonest a LOT of WINE / Proton ratings are. I've seen a ridiculous number of "Sound doesn't work half the time; mouse / keyboard doesn't always register; framerate is less than half what it gets on Windows; textures missing on some objects. GOLD" entries over the years...

When there are two ratings right next to each other with the same distro & kernel & stack and similar hardware, with one claiming Gold and the other Garbage, it doesn't exactly inspire trust - especially when the Garbage one lists out exactly WHY it's Garbage, and the Gold just says "played for 2 mins, works fine". (Though, yeah, obviously there isn't really much TO write if it genuinely does work fine, but...)

Even Silver, at "generally is playable", is a pretty glorified rating for such a low bar. (I mean conceptually: I more than understand how high that bar is on a technical level). "Generally playable" does not count as "Actually Playable" for any gamer I've ever met. It's like being "Generally alive" - it sounds fine in the abstract, but it's a bit of a problem for the "not-generally" portion of the time. :P

(I wonder how the advertised rating for a game is derived. Mode? Mean? Mean of the 5 most recent ratings?)

At least now it looks like we're close to the end of the "Gold on nvidia, Garbage on ATI" - that's certainly a welcome change.

Every little bit helps, but it's also worth remembering that it's the low-hanging fruit that gets picked first. Proton / DXVK / etc HAS done more in one year to make this a reality than WINE would have in another decade, and that's great. But we're not going to see the same pace next year, let alone for multiple years. I think that's the other part that leads to people having (or advocating) unrealistic expectations on the topic.

[arQon]

Incidentally, there's a rather excellent example of the rating system's "accuracy issues" here: https://github.com/ValveSoftware/Proton/issues/2638