feat: add AMD ROCm support for RX 5700 XT (gfx1010 / RDNA 1)#423
Open
PelaGB17 wants to merge 1 commit intorishikanthc:mainfrom
Open
feat: add AMD ROCm support for RX 5700 XT (gfx1010 / RDNA 1)#423PelaGB17 wants to merge 1 commit intorishikanthc:mainfrom
PelaGB17 wants to merge 1 commit intorishikanthc:mainfrom
Conversation
PelaGB17
commented
Mar 10, 2026
PelaGB17
commented
- Dockerfile.rocm: new multi-stage image based on rocm/dev-ubuntu-24.04:6.3.1-complete
- Sets PYTORCH_ROCM_VERSION=6.3 so uv installs ROCm-backed PyTorch wheels
- Sets HSA_OVERRIDE_GFX_VERSION=10.3.0 to map RDNA1 (gfx1010) to RDNA2 kernels
- Adds container user to 'video' and 'render' groups for /dev/kfd + /dev/dri access
- docker-compose.rocm.yml: pre-built image compose file
- Mounts /dev/kfd and /dev/dri (no NVIDIA plugin required)
- group_add: video, render for GPU device node access
- Documents per-generation HSA_OVERRIDE_GFX_VERSION values
- docker-compose.build.rocm.yml: local build variant using Dockerfile.rocm
- base_adapter.go: add GetPyTorchROCmVersion() and update GetPyTorchWheelURL()
- PYTORCH_ROCM_VERSION env var selects ROCm wheel index (rocm6.3, etc.)
- ROCm takes priority over CUDA; falls back to cu126 if neither is set
- All existing adapters (WhisperX, PyAnnote, Voxtral, Parakeet, Canary) inherit ROCm wheel selection automatically via GetPyTorchWheelURL()
- Dockerfile.rocm: new multi-stage image based on rocm/dev-ubuntu-24.04:6.3.1-complete
* Sets PYTORCH_ROCM_VERSION=6.3 so uv installs ROCm-backed PyTorch wheels
* Sets HSA_OVERRIDE_GFX_VERSION=10.3.0 to map RDNA1 (gfx1010) to RDNA2 kernels
* Adds container user to 'video' and 'render' groups for /dev/kfd + /dev/dri access
- docker-compose.rocm.yml: pre-built image compose file
* Mounts /dev/kfd and /dev/dri (no NVIDIA plugin required)
* group_add: video, render for GPU device node access
* Documents per-generation HSA_OVERRIDE_GFX_VERSION values
- docker-compose.build.rocm.yml: local build variant using Dockerfile.rocm
- base_adapter.go: add GetPyTorchROCmVersion() and update GetPyTorchWheelURL()
* PYTORCH_ROCM_VERSION env var selects ROCm wheel index (rocm6.3, etc.)
* ROCm takes priority over CUDA; falls back to cu126 if neither is set
* All existing adapters (WhisperX, PyAnnote, Voxtral, Parakeet, Canary)
inherit ROCm wheel selection automatically via GetPyTorchWheelURL()
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.