v0.10.0 — gfx1151 and gfx90a
One HIP tree, two ISAs. Pass HIP_ARCH yourself; the Makefile does not probe the GPU.
| ISA | GPU | Default DiT | SDPA |
|---|---|---|---|
gfx1151 |
Strix Halo | INT8 + BF16 activations | rocWMMA |
gfx90a |
MI210 | BF16 GEMM | MFMA flash |
Build:
git clone https://github.com/alexhegit/h3-hip.c.git
cd h3-hip.c
git checkout v0.10.0
make HIP_ARCH=gfx1151 -j$(nproc) h3 # or gfx90a| Preset | gfx1151 | gfx90a |
|---|---|---|
| fox-s2 E2E | 83–87 s | ~10.5 s |
| fox-fast E2E | ~95 s | ~18 s |
| 15 s cinematic E2E | 45.0 min | 12 min 33 s |
gfx1151 fox-s2 md5 unchanged from v0.9.0: 1731f95c4aa582597cf83d57f46b8f9e.
Docs: Getting started · Performance