The stuff Hacker News actually recommends.

GPUs and CPUs · NVIDIA GeForce RTX

NVIDIA GeForce RTX 2070 Super

10owners with an opinion
80%of verdicts positive
2/1positive/negative since 2024
6 yrslongest happy owner
1happy after 3+ years

Praised for performance and reliability.

What people wrote

Likes itOwned 6 years
I game on archaic i5-4460 (4 cores, max 3.2ghz), paired with rtx 2070 super and 16gb I can run literally everything new in 50-100 fps range, coupled with 34" 1440p VRR display not much more to desire for single player. Running this maybe 6 years with only graphic card and corresponding psu change. Ie Cyberpunk 2077 everything max apart from rtx stuff and its rarely below 80fps, dlss on good looks. Baldurs gate 3 slightly less but still smooth. God of war same. Usually first think I install is some HD texture pack, so far never any performance hit to speak of. Literally why upgrade? Consoles and their weak performance are effectively throttling PC games last few years (and decade before), not much reason to invest into beefy expensive power hungry noisy setup for most gaming, just to have few extra shiny surface reflections?
Doesn’t like it
I run Steam on Ubuntu with a "GeForce RTX 2070 SUPER" (according to lspci), and while it generally works it has some weird issues with gaming in Linux. Some games end up with what feels like ~200ms latency for no apparent reason, and frame rates on some things like Just Cause 3, which I ought to be horribly overspec'd for (a 2015 game) run comfortably, but just barely, which really isn't right. And Persona 5 gets about 2 frames per second in Linux. My Steam Deck pushes it at 60 at 720p with no problem, and I think was pushing out 1080 at one point quite playably, and I think I benchmarked my PC at ~6 times more powerful than my Steam Deck. Whereas the AMD-based Steam Deck always does what it should do.
Neutral
I would consider 45 fps "solid" for my needs. I rarely saw it dip below 30 for me That said I did see some weird framerates on my AMD system... standing on the bridge of your ship overlooking the planet tanked the framerate hard. It has a 2070 Super and a ryzen 5 3600. But that system is in the living room and used like a game console, and I find Helldivers pretty unplayable with a controller, so I dont use it much
Likes it
Your comment, pre-edit, had something of a severe tone given that consideration. Having said that, I've trained/finetuned image models just fine on an RTX 2070 Super with 8 GB of VRAM. This was back when doing so was more fruitful than simply training a more robust model in the first place. Given that is the current status quo - I'm curious what sort of training you're doing whole-network that actually produces results that are noticeably better than doing something few-shot during inference or doing LoRA finetuning? The latter brings you back into the realm of tuning on low-VRAM configs. In general, a single GPU's memory constraints are one of many when training a model _from scratch_. In that case, you're bottlenecked by data and data parallelism. You don't need one or a few GPU's, you need more than would fit in a consumer setup in the first place.
Neutral
I was playing with this with the llama model was released - and provided I was loading from memory and not disk it was actually only negligbly slower than the model was without sharding (for e.g. the 8gb models that I could load entirely on my GPU). Each layer took a lot more time to evaluate than it took to copy the weights onto the GPU (RTX 2070 super).
Likes it
Both of my LG Blu-ray drives rip 4K discs as easily as standard 1080p discs. My workflow is a script that calls MakeMKV's CLI with the settings I use every time. Then I run that through Handbrake to make a reasonable sized file. Just a handful of clicks. Having a recent-ish (RTX 2070 Super) really helps as the NVenc H264/H265 engine is blazing fast (100+ fps), and is the first hardware encoder I've used that always beats Handbrake/x264 for quality and filesize.
Neutral
Okay I can't try it out anyway. "Blocksparse is not available: the current GPU does not expose Tensor cores" My "best" GPU is an RTX 2070 Super, Turing architecture. I've seen similar messages when using stable-diffusion... either with -webui or with automatic, can't exactly remember, but they both run fine on that RTX 2070 Super, so I can only guess that they revert to some other method than Blocksparse on seeing that it doesn't support Turing. Or something. I haven't looked into how they deal with it. I've submitted an Issue [0] for it. I don't have enough knowledge to know if there's some way of saying "don't use Blocksparse" for fontogen. [0] https://github.com/SerCeMan/fontogen/issues/2
Likes it
I just did a fresh Arch install last week. Running 6.4.1 kernel on a ThreadRipper CPU with an RTX 2070 Super and an RTX 3060. Driver version 535.54.03 and CUDA 12.2. Everything "just works". There was no manual configuration, tweaking or hacking around needed. No issues running Wayland, Proton is handling gaming beautifully and not a sign of screen tearing. The experience is different for everyone it seems.