Edit: The video of that’s the original source of the test results had flawed testing for some of the games and seemingly didn’t set the same tdp limits when switching between the OSs. I’m guessing we’ll see other people test this out later, but for the time being it’s best to assume these results are exaggerated or wrong.
Good catch, results didn’t seem suspicious because Linux has had major performance improvements over windows on previous handhelds.
https://youtu.be/OwWRCrGoXV0?t=284 Digital Foundry tested few games and found no significative differences. But it’s still a big win for linux gaming since those game aren’t native.
I’m also suspicious of those tests since they don’t show details like power usage. Same problem Tom’s did here, took the data at face value missing the clear signs in the original video showing how it was tested improperly.
There are many tweaks and bugs like 17w being 20w+ with this version of the os.
Here’s another video on it that goes more in depth: https://youtu.be/CJXp3UYj50Q
Basically the SteamOS version gets better fps in most games while also have way better battery life. Lower power games would get 2x the battery life when fps when both versions were targeting the same settings/fps.
Basically windows naturally has more overhead than Linux, which really impacts performance/battery life in low power games. But vulkan also typically outperforms directX, so many directX see a noticable performance increase on SteamOS as well (although some amount of this performance gain can be had on windows as well by using dxvk).
I’ll take a look later but quick check he is using two different devices which means its not identical hardware/ firmware.
Also there are many many many more cases where performance has been the same
https://www.reddit.com/r/WindowsOnDeck/comments/1fo7ngs/better_game_performance_on_windows_or_steam_os/
And countless others who see no difference between windows and Linux.
Most times if there is is due to dxvk / Proton fixing a game bug/bottleneck, or invalid testing with not identical settings/hardware.
There is no way windows is using more power and giving lower clock rates without some other bug effecting it.
Edit: finished the video. He didn’t show any actual testing its basically as in depth as the Tom’s article.