Every now and then, the people behind ProtonDB release data dumps from user generated reports of games tested in Steam Play.
What is ProtonDB? It's an unofficial statistics website, where you can note how good or bad various games run on Steam when played with Steam Play. It's a fun one to track, so with a new data dump I thought it might be interesting to try my hand at getting something out of the data for anyone curious.
The latest data dump is available here, which shows a total of 29,823 reports so it's getting rather massive. It goes to show just how many people are testing a wide variety of games!
First up, here's a look at how many reports are going in monthly:
Clearly less reports going into it as time goes on, which is quite interesting. Now, it doesn't mean Steam Play is suddenly any less popular of course, we do have to remember this is an unofficial website and given how many games have a lot of reports already, some may now just look to see existing reports.
Specific to this month, there were 3,119 reports. Here's a breakdown of those reports (one didn't seem to have a status set correctly):
As for the actual games being reported, here's the top 15 most reported games on ProtonDB through January:
Quite an interesting and varied selection there, with Yakuza 0 being no surprise since it was part of last month's Humble Monthly.
It's also nice to see so many people trying out (and enjoying) The Witcher 3, with it being the second most popular game. I'm not entirely surprised here, as I imagine a number of people purchased it originally when a big announcement was made about it coming to SteamOS/Linux which never actually happened in the end. It's another reason why Steam Play is such a great idea, for the times we are sadly let down and for those who don't care about port arguments who just want to play great games.
Here's a look at some more details like Linux Distributions. This one is a close-enough approximation, as the list of distributions is quite varied and messy on the naming, as it's gathered from Steam's system information output:
Note: Some distributions only come up as "Linux 4.x". For those, we're verifying using the Kernel string for both Arch, Fedora and Gentoo. Still looking over ways to do the same for others. All other distributions are bundled into "Other".
Next, here's a look at what types of GPU people are using when attempting to play games with Steam Play:
And again here with what CPU people are using:
Additionally, most reports (1,757 of them) were made using the latest available Proton Beta 3.16-6.
Hopefully I will take another look when they do their next database dump, if people find it interesting enough of course.
Though I don't know if it's included in there; Debian stable is still on Nvidia driver version 390.87, below the minimum recommended value for Proton, so while it accepts my contributions it warns that they won't be shown by default. (I do enjoy the stability of Debian stable, but ugh, graphics driver updates take FOREVER. :()
I don't know if this is included in the data, but it would be interesting to see if there's any difference in the playability of games based on the OS. Like, does the same game run at 'platinum' level more often on Arch-based systems than Debian-based, or vice-versa, or no difference overall? Checked over many different games, of course.
Last edited by Shmerl on 3 February 2019 at 5:37 am UTC
Quoting: GuestWell, I heard a rumor about a year ago that they were using Vulkan for Cyberpunk, but I couldn't find any evidence that it was true. So, I don't believe it.
I don't think there is any evidence of that. CDPR are in general very reluctant to discuss any technical details about their work in progress. So it's possible we won't know what they are using until the very release.
Last edited by Shmerl on 3 February 2019 at 5:49 am UTC
Quoting: NeverthelessFlatpak Steam on any distribution also turns up as "Linux 4.x".I can validate, just had the Steam survey today and saw "Linux 4.x", although I'm on Fedora.
Maybe listing all these as Arch skews up the numbers a little, especially since how much Fedora devs are the ones pushing for Flatpak.
Last edited by Creak on 3 February 2019 at 5:59 am UTC
I have more reports to do..
Last edited by Comandante Ñoñardo on 3 February 2019 at 6:18 am UTC
Quoting: Comandante ÑoñardoThese are my platinum reports:
I have more reports to do..
What's the deal with the Metro games? Do they run better with proton or something?
I would've expected Mint second.
Of course, these stats say nothing about the people that made the report, right?
It would, in theory, be possible that each report is by another person, or that one person is responsible for half the Ubuntu reports.
I wonder how what such a "cleaned up" stat (as in, only one report counts per person) would look like.
Last edited by TheSHEEEP on 3 February 2019 at 7:31 am UTC
Quoting: TheSHEEEPSo many people using Arch?
I would've expected Mint second.
Seems to be in line with GOL stats too: https://www.gamingonlinux.com/users/statistics
Quoting: Comandante ÑoñardoThese are my platinum reports:
I have more reports to do..
See... i dont get why people mark them as platinum when there are so many workaraound involved... thats not platinum rating.
Quoting: MagbedFully agreed.Quoting: Comandante ÑoñardoThese are my platinum reports:
I have more reports to do..
See... i dont get why people mark them as platinum when there are so many workaraound involved... thats not platinum rating.
If you have to do anything else than just launch the game, it is not Platinum. Same if the performance is MUCH worse than on Windows.
If you can make the game run perfectly with some steps, as many of these reports say, that's Gold, Not Platinum.
What we have here is someone unconsciously (I don't think that is on purpose) skewing statistics, and that's not helping anyone in the long run.
The reports and hints are appreciated, but almost none of these are reports that should be Platinum.
I feel that ProtonDB should have functionality to report a report as having the wrong rating.
Last edited by TheSHEEEP on 3 February 2019 at 8:28 am UTC
Quoting: LightsongThere is no way Arch can have such an audience. OpenSUSE for example also gets reported as Linux 4.x
Quoting: CreakGiven that new information, I have adjusted the way that's done. For Arch, Fedora and Gentoo, it's now verifying with the Kernel string too as Arch 99% of the time has arch in the Kernel version string and Fedora 99% of the time has ".fcxx." and Gentoo often does similar.Quoting: NeverthelessFlatpak Steam on any distribution also turns up as "Linux 4.x".I can validate, just had the Steam survey today and saw "Linux 4.x", although I'm on Fedora.
Maybe listing all these as Arch skews up the numbers a little, especially since how much Fedora devs are the ones pushing for Flatpak.
All other smaller distributions are being bundled into Other now too like Sabayon, SteamOS etc. It's a lot more accurate now, but never will be entirely due to Flatpak Steam. Not meant to be an exact science, but it's now accurate enough.
And yes, if you're reporting something as Platinum yet you've used tweaks - that's not Platinum.
Edit again: Added in CPU vendor info.
Last edited by Liam Dawe on 3 February 2019 at 10:00 am UTC
Last edited by callcifer on 3 February 2019 at 9:48 am UTC
That said, there are 1576 open issues, 647 closed.
The open issues that are whitelist requests are 445 (that means games that work without problems, with a Platinum or Gold rating), while 913 issues are "Game Compatibility - unofficial" (that means that the game can work with workaround or not at all, it depends).
So we have already nearly 500 games that are click'n'play, that's great! :D
Quoting: TheSHEEEPQuoting: MagbedFully agreed.Quoting: Comandante ÑoñardoThese are my platinum reports:
I have more reports to do..
See... i dont get why people mark them as platinum when there are so many workaraound involved... thats not platinum rating.
If you have to do anything else than just launch the game, it is not Platinum. Same if the performance is MUCH worse than on Windows.
If you can make the game run perfectly with some steps, as many of these reports say, that's Gold, Not Platinum.
What we have here is someone unconsciously (I don't think that is on purpose) skewing statistics, and that's not helping anyone in the long run.
The reports and hints are appreciated, but almost none of these are reports that should be Platinum.
I feel that ProtonDB should have functionality to report a report as having the wrong rating.
Well first of all, he only notes tweaks for two of his many titles, and one of those is a precaution, not a requirement. Secondly, while I agree that any tweak downgrades it to Gold at best, it's not like you're going to cherry-pick one user's reports to base your opinion on a game.
Indeed, you can't even see a specific user when browsing. So if a few folk are misusing Platinum/Gold ratings, it's still useful overall.
See more from me