I noticed CachyOS’s wiki has a lot of the current relevant envs here: https://wiki.cachyos.org/configuration/gaming/

People complain about performance and then complain about a patch. I’m starting to think social media only cares about outrage.

It’s the JIT shader compilations that need to be rebuilt in the state cache, other games do the same thing it’s why there are tiny stutters the first time you see a new effect in some games.

The argument is that this tech is being used by both the manufacturer and game devs to be lazy and market lies not how can we ever get to 1000hz with path tracing.

Yeah, marketing lies. I mentioned this in the last paragraph.

The whole 500hz benefits are skeptical and subjective at best considering even going from 144 to 240 you’re already seeing diminishing returns on but that’s really a whole other argument about monitor BS currently.

You’re skeptical of the benefits, that is obvious.

You’re wrong about it being subjective though. There are peer reviewed methods of creating photographs that display motion blur as a human eye would experience it. People have been using these techniques to evaluate monitors for years now. Here’s a very high level overview of the state of objective testing: https://blurbusters.com/massive-upgrade-with-120-vs-480-hz-oled-much-more-visible-than-60-vs-120-hz-even-for-office/ . We are seeing diminishing returns because it, roughly, takes a doubling in the refresh rate to cut the motion blur in half. 60-120 is half as blurry, 144 to 240 is only 25% less blurry.

If you want to keep seeing noticeable gains, up to being imperceptible, then display refresh rates need to continue to double and there have to be new frames generated for each of those refresh rates. Even if a card can do 480fps on some limited games, it can’t do 1000fps, or 2000fps.

We need exponential increases in monitor refresh rates in order to achieve improvements in motion blur, but graphics cards have not been making exponential increases in power in quite some time.

Rasterization and Raytracing performance growth is sub-exponential while the requirements for reducing motion blur are exponential. So either monitor companies can decide to stop improving (not likely since TCL just demoed a 4k 1000hz monitor) or there has to be some technological solution for filling the gap.

That technological solution is frame generation.

Unless you know of some other way to introduce exponential growth in processing power (if you did you would win multiple Nobel prizes), then we have to use something that isn’t raw rendering. There is no way for a game to ‘optimize’ its way into having 10x framerate, or 100x framerate.

Being a complex solution doesn’t make it a good solution and frame gen is not a good solution for making sure your game doesn’t run like ass.

Yes, game companies are lazy and they cover the laziness by marketing their game with a lot of upscaling so that they can keep producing crazier and crazier graphics despite graphics cards performance growth not keeping up. This is the fault of gaming companies and their marketing and not of upscaling and frame generation technology

Frame generation is supposed to help older cards get better “FPS” and smooth out motion, you know what would help that over having new games use frame generation as a big ass crutch? Optimizing your damn game so you don’t stutter like a drunken sailor with a speech impediment in the first place and not adding a crap ton of latency with fake frames.

Frame generation gives all cards better FPS, which objectively smooths out motion. Going from 30 to 60 fps cuts motion blur in half. Nothing supposed about it.

A developer’s choice to optimize their game and their choice to support upscaling and frame generation are not mutually exclusive choices. There are plenty of examples of games which run well natively and also support frame generation and upscaling.

Also, frame generation only adds latency when the frame time is long (low FPS). As the source framerate increases the input latency and the frame time converge. In addition, it’s possible to use frame generation to reduce input delay (blur busters: https://blurbusters.com/frame-generation-essentials-interpolation-extrapolation-and-reprojection/). Input latency is a very solvable problem.

My point is that you’re not understanding the trajectory of display hardware development vs the graphics card performance growth and presenting frame generation and upscaling as some plot by game developers and graphics card designers so that they can produce worse products.

It’s conspiracy nonsense.

There’s like 3 rendering engines. Not everything can be described with a tier list.

Frame generation objectively reduces motion blur and frame consistency.

Neural network-based upscaling is a far better alternative. Previously, in the time of the dinosaurs, we’d get better frame rate by turning the resolution down and letting the monitor handle upscaling. This looked bad but higher frame rate often is more important for image quality than resolution. Now we get the same performance boost with much less loss of visual clarity, and some antialiasing for free on top of it.

Upscaling and frame generation are good technologies. People are upset at the marketing of graphics cards which abuse these technologies to announce impressive FPS numbers when the hardware isn’t as big of an upgrade as implied.

Marketing departments lying about their products isn’t new, but for some people this is the first time that they’ve noticed it affecting them. Instead of getting mad at companies for lying, they’re ignorantly attacking the technologies themselves.

Frame generation is a requirement if we’re going to see very high refresh rate (480hz+) displays become the norm. No card is rasterizing an entire scene 500 times per second.

Calling it fake frames is letting Internet memes stand in place of actual knowledge. There’s a lot of optimizations done in the rendering pipeline which use data from previous frames to generate future frames, generating an intermediate frame while waiting for the GPU to finish rendering the previous frame is just one trick.

The generated frame increases the visual clarity of motion, you can see at https://testufo.com/photo.

We’re not going to have cards that can pathtrace at 4k@1000hz anytime soon, frame generation is one of the techniques that will make it possible.

It’s one thing to be upset at companies marketing teams who try to confuse people with FPS numbers by tweaking up scaling and frame generation. Directing that frustration at the technology itself is silly.

e: a downvote, great argument

No matter how optimized a game is, there will be someone with hardware that can barely run it.

For those people, having access to upscaling in order to gain performance is a plus.

Honestly, I’d bet money that the problem is at least partly Denuvo‘s fault.

So many envs and no central documentation makes it hard to handle for everyone.

Ah, I didn’t RTFM completely. I just read a snippet that mentioned the NGX updater and misunderstood the context.

I’ll give it a shot, thanks.

it’s there on others (edges can get a bit shimmery with tsr) but really bad with dlss

Yeah that’s my experience as well. TSR seems to be doing the same thing but it isn’t applying the oversharpening which makes them stand out.

What launch options are you using if you don’t mind?

ENABLE_HDR_WSI=1 PROTON_ENABLE_HDR=1 PROTON_ENABLE_WAYLAND=1 gamemoderun %command%

You have to be using GE-Proton10 or above in order to use Wayland’s HDR. I don’t think you need ENABLE_HDR_WSI (I believe PROTON_ENABLE_HDR makes Proton set a bunch of environmental variables on startup.) but I’m not sure.

Using GE-Proton10-15, HDR works great too.

I did notice the edge flickering artifacts with upscaling. XeSS is a bit higher quality than TSR but it also has the flickering. FSR framegen causes the flickering to happen on some particle effect that they use for atmosphere effects (like pieces of dust floating in the air) so it isn’t very usable currerntly.

The game isn’t perfect, but it’s very playable for me after some settings adjustments. I didn’t have any crashes in 5.5 hours of playtime, but I did notice the shader compiling stutter and there were some spots where you could tell that it was loading a zone if you walked over a specific point and I was in combat at the time so I ran across that point a few times and that caused some framerate issues.

A HUGE amount of the stuttering was eliminated by setting the Textures Streaming Speed to Very High, it looks like this is throttling disk IO for performance reasons. If you have an NVME SSD then I can’t think of a reason not to set it to reason not to set it to very high.

That’s good to know that Steam co-op works. I’ll try it later today, my friends are all running Linux too and didn’t want to buy a copy if it wasn’t going to work. I happened to be home yesterday so I was the guinea pig.

I tried updating the DLSS version (using PROTON_ENABLE_NGX_UPDATER=1), the flickering still occurs. Same with using RENDER_PRESET_K. It almost looks like they’re applying too much sharpening when you’re using DLSS, but I don’t see a way to adjust that specifically.

Playing, on Linux (Arch, btw) with no issues. The defaults were a bit harsh (45fps@4k) but once I ran the graphics setting autodetection and it went to medium with balanced upscaling I was getting 60+ FPS.

I couldn’t connect to the matchmaking servers (my system doesn’t meet the requirements, apparently) but it otherwise ran just fine.

Are you playing on an 8K display or something?

https://github.com/alexghergh/nvim-tmux-navigation

You want something like that. It makes navigating between the two applications use the same set of keys, there are some other plugins (linked in the readme) that do similar things with different feature sets, but fixing the keybinds is a huge step towards making it a smooth experience.

It seems a bit dated too, I definitely have a bitlocker disk mounted right now.

The TPM secure boot key part is true, but you can disable secure boot and use (on Windows) manage-bde to add a password based key. When you boot and it can’t load the bitlocker key from the TPM it prompts the user for a password.

I’m fairly sure there’s a way to bypass the secure boot requirement for Windows 11 too, I think I read about it but I’m not using windows so I didn’t look into it much.

It’s a pain. Windows can’t read ext4 and NTFS does not play well with Proton gaming. You could use exfat as a common filesystem. My solution before I gave up duel booting was to put games on my NAS and access them via NFS. It was a bit slow even on a 2.5gb network, but very playable.

Yeah, staying on top of updates and watching the announcements is pretty much a requirement for Arch. Things change fast, that’s both the advantage and disadvantage of this distro.

The AUR, like any repository where users can submit software, is going to have malicious uploads from time to time. The AUR team does a good job of removing these as soon as they’re discovered but there’s nothing that can prevent it. There’s a voting system so you usually see which packages are the most commonly installed but even that could be manipulated if someone were motivated enough.

I would guess that if it became common enough they could enable some more stringent identity verification for submitters in order to cut down on bad actors. But it is very much at your own risk and there’s a big warning about it in the wiki saying as much.

I was just guessing (it’s how I’d do it) 🥳