# AMD ROCm 47 Hacker News commenters recommended AMD ROCm and 97 criticized it, November 2025 to November 2026. It was recommended in 11 of 13 months; owners who said how long report a median of 3 years (4 owners). **Verdict:** AMD's open-source GPU compute stack for running AI models and CUDA-style code on AMD cards; support has improved a lot, but newer chips and consumer cards often lag behind. ## Why people recommend it - Support has improved markedly recently (6 comments): "The "ROCm" situation with Strix Halo was pretty bad for a while. I think it finally stabilized late last year." ([source](https://news.ycombinator.com/item?id=48813203)) "around 7.2 it became reliable and as fast as or faster than ROCm 6.4." ([source](https://news.ycombinator.com/item?id=47825955)) "It's a LOT less bad than it used to be, amd deserves serious credit." ([source](https://news.ycombinator.com/item?id=47056322)) - Runs most AI models and frameworks (6 comments): "Been able to run all models on day 0 at comparable speeds to Nvidia with similar memory bandwidth." ([source](https://news.ycombinator.com/item?id=49886865)) "Yes, ROCm can be used to run frontier models and is being used by OpenAI, Anthropic, and Meta." ([source](https://news.ycombinator.com/item?id=49173059)) "ROCM can be installed from the regular repo, and most AI software has ROCM builds by now." ([source](https://news.ycombinator.com/item?id=49290237)) - AMD backing and production use (3 comments): "it's first party supported by AMD (ROCm LLVM, with a lot upstreamed as well) where it's fairly widely used in production" ([source](https://news.ycombinator.com/item?id=47056172)) "Yes, ROCm can be used to run frontier models and is being used by OpenAI, Anthropic, and Meta." ([source](https://news.ycombinator.com/item?id=49173059)) "they're actually trying to invest in ROCm and their Instinct ecosystem" ([source](https://news.ycombinator.com/item?id=45933833)) - Packages for Windows, Linux and Debian (3 comments): "ROCM can be installed from the regular repo, and most AI software has ROCM builds by now." ([source](https://news.ycombinator.com/item?id=49290237)) "AMD's drivers actually come with a recent version of torch that 'just works' on windows" ([source](https://news.ycombinator.com/item?id=47051142)) "Debian build their ROCm with support for all possible devices." ([source](https://news.ycombinator.com/item?id=47813212)) ## What people warn about - Lags on newer and consumer GPUs (5 comments): "ROCm is up and coming, but support ... is still a work in progress, especially for newer chips." ([source](https://news.ycombinator.com/item?id=47258827)) "Supporting only Server grade hardware and ignoring laptop/consumer grade GPU/APU for ROCm was a terrible strategical mistake." ([source](https://news.ycombinator.com/item?id=47749703)) "The day ROCm supports EVERY AMD card on release, just like CUDA does, is the day I will actually believe this marketing hype." ([source](https://news.ycombinator.com/item?id=47750349)) - Crashes and breakage after updates (4 comments): "ROCm works great too, the only issue i have had is that my machine froze a couple of times" ([source](https://news.ycombinator.com/item?id=47932031)) "imo it's been a buggy mess, and I'm talking bugs that make your entire system lock up until you hard reboot" ([source](https://news.ycombinator.com/item?id=46890707)) "it is way less reliable; any upgrade in kernel version or AMD driver and suddenly everything is broken" ([source](https://news.ycombinator.com/item?id=47824179)) - Hard setup and manual patching (4 comments): "ROCm stack is not for people though who aren't willing to dig in and patch things themselves." ([source](https://news.ycombinator.com/item?id=48563510)) "It has been a bit of a nightmare and had to package like 30+ deps" ([source](https://news.ycombinator.com/item?id=47745792)) "I guess it counts if you can find the information from one of the many conflicting wikis out there and then figure out how to hack support for your card" ([source](https://news.ycombinator.com/item?id=47750988)) - Slow bug fixes from AMD (3 comments): "I have been trying to use Jax on AMD/ROCM since 2023 and still can't because of the countless ROCM bugs and terrible response time of their team." ([source](https://news.ycombinator.com/item?id=47795744)) "ROCm didn't have such a team until hyperscalers lost their shit over our visibly poor development practices." ([source](https://news.ycombinator.com/item?id=45933511)) "how AMD and Intel keep failing to understand what CUDA, the whole ecosystem, is actually about" ([source](https://news.ycombinator.com/item?id=47749861)) ## Alternatives hackers recommend - [NVIDIA CUDA](https://hackerapproved.com/p/cuda.md): 91 recommend - [BarraCUDA](https://hackerapproved.com/p/barracuda.md): 21 recommend - [Triton](https://hackerapproved.com/p/triton.md): 18 recommend - [ZLUDA](https://hackerapproved.com/p/zluda.md): 13 recommend - [HIP (AMD ROCm)](https://hackerapproved.com/p/hip.md): 12 recommend ## Cite as Hacker Approved, "AMD ROCm", https://hackerapproved.com/p/rocm/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).