ROCm
Fewer than 5 people recommend it so far, so there's no written verdict. Here is what commenters said.
Why people recommend it
“ROCm runs most things just fine without any ceremony or difficulty.”
“Of course, it's not done at all, and that looks bad, and I'm so scared to share it, in that light, but I just want to applaud so much”
Show 2 more
“I'm able to use ROCm with MTP just fine (Qwen and Gemma).”
“Now that ROCm is reliable, I'm using ROCm.”
Caveats and dissent
“The official llama.cpp ROCm build detected the card correctly, but kept crashing with 0xC0000005 while loading the model”
“ROCm, to this day, still sucks.”
Show 2 more
“It definitely wasn't an "out of the box" experience, fiddling around with kernel versions and ROCm installs.”
“This is about the GPU clock remaining at max frequency, but that can drive the fan speed to increase.”
Often compared with
Who recommends it in their own words
Threads where it came up
- CUDA for AMD on Windows 1 mention
- Benchmarking Qwen3.8 27B quantizations: 4-bit holds up, 1-bit collapses 1 mention
- We Have AI at Home Chapter 1: A Box of Scraps 1 mention
- Can AMD Break the CUDA Moat? AMD Advancing AI 2026 1 mention
- AMD's Instinct MI455X: Aiming for the Sun 1 mention
- New Framework Desktop Option with AMD Ryzen AI Max+ Pro 495 and 192GB Memory 1 mention
Data: rocm.json. Quotes are verbatim; the verdict and theme labels are written by a model from these comments, and every clause links to its sources.
All comments 10
“using ZLUDA + ROCm/HIP”
“The official llama.cpp ROCm build detected the card correctly, but kept crashing with 0xC0000005 while loading the model”
“ROCm runs most things just fine without any ceremony or difficulty.”
“Of course, it's not done at all, and that looks bad, and I'm so scared to share it, in that light, but I just want to applaud so much”
“ROCm, to this day, still sucks.”
“I'm able to use ROCm with MTP just fine (Qwen and Gemma).”
“It definitely wasn't an "out of the box" experience, fiddling around with kernel versions and ROCm installs.”
“ROCm is finally usable.”
“This is about the GPU clock remaining at max frequency, but that can drive the fan speed to increase.”
“Now that ROCm is reliable, I'm using ROCm.”