Why am I getting lower FPS on Linux with the same RTX 4080 Super?

November 20, 2026 4 answers
Asked by lootPouch+Radish

I'm comparing Windows and CachyOS on a Ryzen 7 7700X, RTX 4080 Super and 32 GB of RAM, using a 4K monitor. Cyberpunk 2077 gets around 96 FPS on Windows versus 73 on Linux with DLSS Performance, path tracing and frame generation. Crimson Desert also runs slower on Linux.

GPU usage is near 99% on both, but Linux draws less power. Different Proton versions haven't made much difference. What should I check?

Answers

QTEthumbFumble_4E

First, check whether frame generation is actually active on both systems. The displayed frame rate can include generated frames, so a mismatch there would make the comparison pretty misleading. Check the game settings after each launch, not just the preset name.

Then try the same repeatable scene with frame generation off on both systems. Match resolution, DLSS mode, path tracing and other graphics settings, and compare both the average and the dips. That should tell you whether the gap is in the underlying rendering performance or mainly in the generated-frame setup.

lootSlot+Kettle:

Also make sure both FPS counters are measuring the same thing. A counter may not report generated frames the same way as another overlay.

Reply to this answer
laggingLegend

Near 99% GPU usage doesn't rule out a graphics stack bottleneck. On Linux, Windows games generally go through Proton and a graphics translation layer, and ray tracing workloads can behave differently across drivers and APIs. The lower power draw could be a symptom of the GPU waiting on parts of that work rather than proof of a power limit.

Check the NVIDIA driver version and whether the driver is loaded correctly, then compare a run with path tracing off. If the gap shrinks a lot, that gives you a useful direction without assuming the cause is Proton itself. Changing Proton builds alone won't necessarily change the driver or every part of the rendering stack.

Reply to this answer
quickscopeByAccident

I'd check the boring stuff too. Confirm the monitor is actually running at the intended resolution and refresh rate in both systems, and make sure there isn't a frame cap or sync setting enabled on only one. Overlays and capture tools are worth disabling for a test as well.

Reply to this answer
mechLoadoutJuggler

Use a fixed benchmark or the same short route for each run. Combat is useful for finding bad dips, but it's hard to reproduce exactly. Run each test more than once after shaders have had a chance to compile, and note the driver, game and Proton versions alongside the results.

For Crimson Desert, do a separate comparison with matched settings rather than assuming both games have the same problem. A gap that persists across repeatable tests is much easier to troubleshoot than a single FPS reading.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions