Commit 2e1aba6
Add FBGEMM 1.6/1.7/1.8 entries to PyTorch compat table (#5941)
Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/2857
Pull Request resolved: #5941
The `_fbgemm_torch_compat_table` in `fbgemm_gpu/__init__.py` lagged behind the published FBGEMM-PyTorch compatibility table. It topped out at FBGEMM `1.5` -> PyTorch `2.10`, while `Releases.rst` already lists `1.6.0` -> `2.11.x`.
This adds three entries at the top of the table:
- `1.6` -> `2.11` (confirmed against `Releases.rst`)
- `1.7` -> `2.12` (confirmed by the v1.7.0 release announcement: https://fb.workplace.com/groups/fbgemmusers/permalink/27222344027419687/)
- `1.8` -> `2.13` (inferred from the minor-for-minor pattern)
> **NOTE:** `1.6` is backed by the published `Releases.rst` table and `1.7` by the v1.7.0 release announcement, but neither `1.7` nor `1.8` is yet reflected in `Releases.rst` (the doc still tops out at `1.6.0`). The `1.8` mapping is extrapolated from the consistent minor-for-minor PyTorch versioning pattern. These rows should be reconciled with `Releases.rst` once the corresponding FBGEMM releases are cut and their actual PyTorch / Python / CUDA / ROCm support matrices are known.
Reviewed By: spcyppt
Differential Revision: D109308705
fbshipit-source-id: 085e823fd28f701eb5d62d6f2f855e2fbe3c5eb11 parent 17dd8e3 commit 2e1aba6
1 file changed
Lines changed: 3 additions & 0 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
15 | 15 | | |
16 | 16 | | |
17 | 17 | | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
18 | 21 | | |
19 | 22 | | |
20 | 23 | | |
| |||
0 commit comments