The stuff Hacker News actually recommends.

GPUs and CPUs · AMD Radeon RX 7900 XTX

AMD Radeon RX 7900 XTX 7900

10owners with an opinion
57%of verdicts positive
4/3positive/negative since 2024
0happy after 3+ years

Praised for performance and reliability; criticized for software.

What owners wrote

Neutral
I ran the Strata project with this on 32gb ram 16gb AMD 7900XTX. It goes about 20tok/s off the ssd. I'm guessing 64gb would go to 40tok/s. https://github.com/Niko1221/Strata Quality wise its not fast enough for me to eval. I'll probably spend some tokens on it through openrouter. DeepSeek 4 flash though is a mess compared to GPT Sol 6.1.
Likes it
I had a 6950 on my pc from when I built the thing... and then bought the 7900 for $5xx, that allows me to run more models, and then I saw the "Radeon AI PRO" and after a couple of frustrating talks with certain LLM to try to get an idea on what the speed of the card is I decided to go, buy it and test it to check what's the actual speed.
Neutral
Been running local LLMs on my 7900 XTX for months and the ROCm experience has been... rough. The fact that AMD is backing an official inference server that handles the driver/dependency maze is huge. My biggest question is NPU support - has anyone actually gotten meaningful throughput from the Ryzen AI NPU vs just using the dGPU? In my testing the NPU was mostly a bottleneck for anything beyond tiny models.
Likes it
Agreed. I was sorely tempted by the Mac studio, but ended up with a 96GB ram Ryzen 7900 (12 core) + Radeon 7800 XT (16GB vram). It was a fraction of the price and easy to add storage. The Mac M2 studio was tempting, but wasn't refreshed for the M3 generation. It really bothered me that the storage was A) expensive, B) proprietary, C) tightly controlled, and D) you can't boot without internal storage. Even moving storage between Apple studios can be iffy. Would I be able to replace the storage if it died in 5 years? Or expand it? As tempting as the size, efficiency, and bandwidth were I just couldn't justify top $ without knowing how long it would be useful. Sad they just didn't add two NVMe ports or make some kind of raw storage (NVMe flash, but without the smarts).
Likes it
If you put your foot to the floor in a supercharged car, you're going to eventually have to let it up lest you melt things or you burn all your oil because your rings aren't making contact with the cylinder walls any more. It's an apt metaphor since the same is true of CPUs. You can't run a CPU at 400 watts continuously for more than a handful of seconds at a time. The problem is that Intel has normalized it so much that all their high end CPUs do this, and apparently do it often. It's not unexpected that they might be too close to the point where things are melting, so to speak. I'd rather slower and more stable any day - I chose a Ryzen 7900 over a 7900X intentionally - but that isn't what all the marketing out there is trying to sell. The fancy motherboards, the water coolers, the highly clocked memory all account for lots of markup, so that's what's marketed. I'm not a fan. It is worth noting a distinction between the terms "overclocking" and "turbo clocking". "Overclocking" has traditionally meant running the clock "over" the rating. "Turbo clocking" is now built in to almost every CPU out there. One technically can void your warranty, whereas the other doesn't. Since we're mostly technical people here, we should use the appropriate term where the context makes that choice more accurate. It's like virus and Trojan - we SHOULD be technically correct, but that doesn't mean highly technical people aren't still calling Trojans viruses now and then.
Doesn’t like it
I'm really rooting for AMD to break the CUDA monopoly. To this end, I genuinely don't know whether a translation layer is a good thing or not. On the upside it makes the hardware much more viable instantly and will boost adoption, on the downside you run the risk that devs will never support ROCm, because you can just use the translation layer. I think this is essentially the same situation as Proton+DXVK for Linux gaming. I think that that is a net positive for Linux, but I'm less sure about this. Getting good performance out of GPU compute requires much more tuning to the concrete architecture, which I'm afraid devs just won't do for AMD GPUs through this layer, always leaving them behind their Nvidia counterparts. However, AMD desperately needs to do something. Story time: On the weekend I wanted to play around with Stable Diffusion. Why pay for cloud compute, when I have a powerful GPU at home, I thought. Said GPU is a 7900 XTX, i.e. the most powerful consumer card from AMD at this time. Only very few AMD GPUs are supported by ROCm at this time, but mine is, thankfully. So, how hard could it possibly to get Stable Diffusion running on my GPU? Hard. I don't think my problems were actually caused by AMD: I had ROCm installed and my card recognized by rocminfo in a matter of minutes. But the whole ML world is so focused on Nvidia that it took me ages to get a working installation of pytorch and friends. The InvokeAI installer, for example, asks if you want to use CUDA or ROC…
Likes it
Not sure what's going on, but my Ubuntu 23.10 system with a Ryzen 7900 (non-x) with 12 cores is 3.1x faster than the AMD EPYC 9R14 with 64 cores and 2.6x faster than the Intel Xeon 8488C with 48 cores. I just did the usual git clone, mkdir build, and: cmake .. -DCMAKE_BUILD_TYPE=Release -DSIMDJSON_JUST_LIBRARY=OFF To get the benchmark results: ~/git/simdjson/build/benchmark$ cat log | grep partial_tweets | grep "simdjson_ondemand" partial_tweets /manual_time 66705 ns 75966 ns 10453 best_bytes_per_sec=9.59662G best_docs_per_sec=15.1962k best_items_per_sec=1.51962M bytes=631.515k bytes_per_second=8.81706G/s docs_per_sec=14.9913k/s items=100 items_per_second=1.49913M/s [BEST: throughput= 9.60 GB/s doc_throughput= 15196 docs/s items= 100 avg_time= 66705 ns]