The stuff Hacker News actually recommends.

GPUs and CPUs

AMD Radeon RX vs Nvidia GeForce

Owners are split: the signals below point different ways.

AMD Radeon RXNvidia GeForce
Owners with an opinion119305
Positive verdicts72% of 8874% of 217
Positive/negative since 202438/1185/30
Happy after 3+ years211
Longest happy owner6 yrs21 yrs
Praised forperformance and reliabilityperformance and reliability
Criticized forsoftwaresoftware
Newest comment

Updated

What owners wrote about both

AMD Radeon RX: Likes itNvidia GeForce: Doesn’t like it
I moved to Linux from Windows 6 months ago. Here are some minor things that always went sour on Windows: Webcam's settings would always get reset semi-randomly The date from the clock would frequently disappear Mic's volume would often get lowered Other than that, with Windows 10 Pro it was super solid on 12 year old hardware. Never crashed, all games I could play were fine and I did heavy Docker based development, highly leveraging WSL 2. Also recorded tons of videos (screencasts with a webcam). Switched to native Linux. Had a rough start but it improved after going down a few rabbit holes. My GeForce 750 Ti (2 GB) didn't play nice with Wayland. The official NVIDIA drivers have big problems allocating system memory when GPU memory is filled causing apps to crash constantly[0]. Games were lagging and stuttering. My whole machine would pause for 5-10 seconds and then unpause, a few times a day. Ended up having to modify a udev setting on my SSD to enable max_performance for its power management and the machine never paused again. Switched to an AMD RX 480 (8 GB) GPU and all of the GPU related problems went away, and the games I played stopped stuttering and lagging. Now it's quite stable but there's still 3 problems. I've gone down pretty deep rabbit holes for each one but didn't resolve them. If I'm downloading a file with Firefox and I start a Docker container that involves configuring a Docker network, my file download will get interrupted. Not 100% of the…
AMD Radeon RX: Likes itNvidia GeForce: Doesn’t like it
Yep I had a GeForce 750 Ti (2 GB) and I was able to run a ton of things on Windows without any issues at all. As soon as I switched to Linux I had all sorts of problems on Wayland where as soon as that 2 GB was reached, apps would segfault or act in their own unique ways (opening empty windows) when no GPU memory was available to allocate. Turns out this is a problem with NVIDIA on Wayland. On X, NVIDIA's drivers act more like Windows. AMD's Linux drivers act more like Windows out of the box on both Wayland and X. System memory gets used when VRAM is full. I know this because I got tired of being unable to use my system after opening 3 browser tabs and a few terminals on Wayland so I bought an AMD RX 480 with 8 GB on eBay. You could say my cost of running Linux on the desktop was $80 + shipping. A few months ago I wrote a long post going over some of these details at https://nickjanetakis.com/blog/gpu-memory-allocation-bugs-wi... . It even includes videos showing what it's like opening apps both on Wayland and X with that NVIDIA card.
AMD Radeon RX: Likes itNvidia GeForce: Neutral
I can say that number is at least 1. I enjoy gaming, but I have been fully on Linux for a few years now. My PC builds always need high CPU performance, both multi threaded for compile workloads and single threaded for games, as well as good graphics performance for gaming, especially due to VR games. I am essentially the reverse of your described use case: I want high gaming performance with ML and GPGPU as a bonus. I like what I am seeing in regards to the new AMD cards vs. my current RTX 3080 for gaming performance. Especially since Nvidia has raised prices to a point I am no longer even begrudgingly willing to pay. But, as I am currently playing around a bit with stable diffusion I am more inclined to stay with CUDA for the time being. However this is a very recent change in interests, so I am not sure whether my priorities will change long term. I am also not planning an upgrade any time soon, so I will wait for new developments on all sides.
AMD Radeon RX: Likes itNvidia GeForce: Neutral
Yes, your GPU needs to support the newest DisplayPort 1.4 standard (with updated DSC) for 4k@144Hz. AFAIK all Nvidia 20xx series cards and Radeon RX 6xxx cards support that. The fallback is 4k@120Hz which should work with older cards, but when testing with my GTX 1070 I had to add a custom resolution to get that set. With my new RX 6900 XT there are no problems.
AMD Radeon RX: Likes itNvidia GeForce: Likes it
I run vanilla Arch with niri (Wayland), Waybar and a few other tools to create a desktop environment. I didn't do any special kernel or app specific tweaks for performance. The machine boots using 1.1 GB of memory and total disk space after installing a bunch of apps is ~10 GB. Browser, Docker, video editor, image editor, LibreOffice, OBS, VMs, a ton of CLI tools and everything you'd expect to have a fully working and usable system is there. It's a wonderful environment on 2014 era hardware. Granted I have 16 GB of memory but it would be fine with 8 GB even when running non-trivial Docker based projects and doing 1080p video editing. Even Silksong runs at 60 FPS with a GeForce 750 Ti (2 GB) GPU. Although I splurged on a used AMD RX 480 (8 GB) to get more things to run nicely in parallel. Wayland is very picky with limited GPU VRAM, especially with NVIDIA cards. I also run the same set up on a more modern laptop with way better hardware but things aren't much faster unless I'm very CPU bound. Everything to get the system up and running in 1 command and 10-15 minutes is here: https://github.com/nickjj/dotfriedrice
AMD Radeon RX: Likes itNvidia GeForce: Likes it
> there are many control circuits and a large number of registers associated with each ALU. Therefore the real percentage of area occupied by multipliers must be many times less than this. Might be true, but I don’t see how that follows necessarily, if every DP multiplier costs as much as 64 logic or add operations, right? Even in single precision, multipliers are much larger than other units. Multiplying by 4 only amplifies that and makes the ratio bigger. > Under this assumption, removing all support for DP would save about 16% of the area You lost me there- if we’re talking about a fixed number of multipliers reduced from double to single precision, and the double multipliers are 32% of the area, then because double is 4x the area, as you pointed out, the savings would be 24% not 16%, right? In other words, a 32% area for N double precision multipliers would take only 8% area for N single precision multipliers, no? > the real savings must be much less Could be true, but I’m not convinced yet, I don’t find this compelling. DP registers are double the size. If you want the same cache hit rates, the cache needs to be double the size. The buses are 2x, all of the math is at least 2x, and multiplies are 4x. GPUs are devoted to math, and double precision takes twice the memory and twice the bandwidth to maintain the same number of ops throughput, so it’s not clear why the area assumption shouldn’t start north of 2x - double as a baseline and a bit higher due to multipliers. > I…