ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Installer: enable ROCm torch on RDNA2 (gfx1030-1036) on Windows (#7277)

Browse source 
* Installer: enable ROCm torch on RDNA2 (gfx1030-1036) on Windows

repo.amd.com publishes a gfx103X-all wheel family with win_amd64 torch
2.9.1/2.10.0/2.11.0+rocm7.13.0 (cp310-313), but both Windows allowlists
omitted RDNA2, so RX 6000 cards (gfx1030/1032, etc.) fell back to CPU-only
torch. Map gfx1030-1036 to gfx103X-all in install.ps1 ($archFamilyMap) and
install_python_stack.py (_GFX_TO_AMD_INDEX_ARCH). No torch floor (mirrors
gfx110X-all: newest wheel, no _grouped_mm bug on RDNA2). NVIDIA/Mac/CPU and
Linux paths untouched; gfx906 stays CPU (no wheels published).

* Sync studio/setup.ps1 RDNA2 (gfx1030-1036) allowlists for PR #7277
This commit is contained in:
Daniel Han 2026-07-21 03:54:25 -07:00 • committed by GitHub
commit 35f887d795
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 17 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

7
studio/install_python_stack.py
View file

@ -352,6 +352,13 @@ _GFX_TO_AMD_INDEX_ARCH: dict[str, str] = {
"gfx1102": "gfx110X-all", # RDNA 3
"gfx1101": "gfx110X-all",
"gfx1100": "gfx110X-all",
"gfx1036": "gfx103X-all",
"gfx1035": "gfx103X-all", # RDNA 2 (RX 6000)
"gfx1034": "gfx103X-all",
"gfx1033": "gfx103X-all",
"gfx1032": "gfx103X-all",
"gfx1031": "gfx103X-all",
"gfx1030": "gfx103X-all",
"gfx90a": "gfx90a",
"gfx908": "gfx908", # MI200/MI100
}