What Kind of Gamers Contribute to ProtonDB?

By

If you are following the Linux Gaming market closely you are probably aware of ProtonDB by now. This site lets Linux users contribute reports regarding the state of Windows games running on Linux via SteamPlay. There’s a wealth of reports and since there is handy dump of the database (without personal information) on github, I had a quick look and we can derive some learnings from there.

While I would certainly like to generalize the below stats to the larger group of Linux gamers, this would probably not work because I expect:

  • People who know and use Proton to be very proactive gamers
  • People who contribute reports to be different from your average user

Data is available since August but ProtonDB recently decided to parse system data from Steam System Info directly, which makes for a better and more accurate system and OS reporting. For this reason I focused on the November 2018 data only.

While there is no way to identify who is who, I took the assumption that any unique combination of OS, RAM amount, CPU model and GPU model would likely be from the same individual. It would be extremely rare that two individuals have the exact same hardware and distro, and I also consider it unlikely that within the same month someone would switch distro/hardware as they provide new reports. Still, these things can happen and therefore the underlying assumption can not be considered 100% accurate. This said, I end up with an estimated n size of 1638 unique contributors, for a set of 4441 reports captured in November 2018.

With the above caveat in mind, I decided to look at a couple of metrics:

  • What are the distros they used and what is their split?
  • What is the share between AMD CPUs and Intel CPUs right now?
  • What is the typical amount of RAM used?
  • What is the proportion of AMD/Nvidia/Intel GPU users?
  • For Nvidia users, what are the most used GPU cards?
  • For AMD users, same question.
  • Who were the most helpful contributors?

Distro-wise, nothing too surprising. This seems pretty consistent with what we observed before: Ubuntu being at 40% or so, followed by Arch-based distributions (Arch, Manjaro, Antergos).

Manjaro seems to be pretty high though, much higher than we have ever seen it before. Among ProtonDB contributors we do see a lot of Solus users (4%) while Solus was only between 1% and 2% in previous surveys. Last but not least, SteamOS usage is very low, but this could be a bias of the Proton effect. SteamOS is certainly not the easiest distro to use Steam Play/Proton with, so we can expect a lower representation of such users there… while my guess is also that SteamOS’s share is declining overall.

Let’s look at hardware components.

It seems like the days of 8GB RAM are pretty much over. At least a majority of contributors are already at 16 GB and above. I am personally still at 8GB on my main gaming machine, but it’s certainly something to consider as more recent ports from Feral for example recommend 16GB or RAM for optimal performance.

CPU-wise, Intel is still by far the leader, but AMD seems to be at a all-time high. This is also something we see in the Steam hardware survey in the CPU page, but even more so among the ProtonDB contributors.

GPU-wise, the current numbers on Steam show that the split is 74.1 % NVIDIA, 15.2 % AMD, and 10.5 % Intel. Among our ProtonDB contributors, we can see that AMD is much, much stronger than the average, and even above our previous surveys.

This should give credence to the fact that AMD’s efforts in terms of driver support in Mesa are paying off.

Looking at NVIDIA users, the landscape has changed quite a lot compared to 2 years back. The GTX970 used to be the most popular card at the time, and now a majority of contributors have moved to the 1000 series, with the GTX 1060 and 1070 being the most popular models. Share wise things are trickling down nicely from one model to the next in terms of share.

The situation among AMD users is quite different. 3 Models represent a large chunk (more than 30%) of the share, and then everything is heavily fragmented. Since I am not too familiar with the current AMD line-up, I cannot come up with a good hypothesis as to why this is happening among AMD users. Dear readers, if you do know what’s happening, please shed some light with us in the comments.

Last but not least, I could also derive that the MVP in ProtonDB is someone who has contributed no less than 123 reports in November alone. At first, it seems doubtful but by looking at the free text comments we can clearly see it’s the same person because the same style of writing is used over and over again. He/She has a GTX1080-based hardware, and reported on 122 different games.

Daily Times (Los Angeles TZ) when our MVP contributes on ProtonDB.

Based on the timing of the report, we can derive two things:

  • This person is located somewhere close to the West Coast (Silicon Valley maybe?)
  • This person has a full time job and posts mainly new reports in the evenings (after 6pm and until after midnight) and on weekends.

I can’t say I noticed a particular pattern in terms of tastes among the titles they tested - it’s pretty eclectic while there is a good amount of AAA games on a quick look.

Finally, the ProtonDB contributors focused mainly on the following 20 games in November, by the number of reports:

RankGameReports
1Grand Theft Auto V48
2The Witcher® 3: Wild Hunt44
3Age of Empires II HD38
4HITMAN™ 231
5DARK SOULS™ III28
6DOOM27
7The Elder Scrolls V: Skyrim26
8No Man's Sky20
9Killing Floor 219
10Warframe19
11METAL GEAR SOLID V: THE PHANTOM PAIN18
12Quake Champions18
13The Awesome Adventures of Captain Spirit18
14Path of Exile17
15Subnautica17
16Fallout: New Vegas16
17Frostpunk16
18Divinity: Original Sin 2 - Definitive Edition15
19Fallout 415
20Sins of a Solar Empire: Rebellion®15

That’s it for this time. Happy New Year 2019 to all!