chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[Diffusion] [NPU] [Doc] Add NPU documentation for sglang-diffusion (#18894)

Browse Source 
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
This commit is contained in:
Makcum888e
2026-02-17 10:12:20 +03:00
committed by GitHub
parent 899e2be7d0
commit 14c95d255c
3 changed files with 17 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/diffusion/index.md
View File

@@ -7,7 +7,7 @@ SGLang Diffusion is an inference framework for accelerated image and video gener
- **Broad Model Support**: Wan series, FastWan series, Hunyuan, Qwen-Image, Qwen-Image-Edit, Flux, Z-Image, GLM-Image, and more
- **Fast Inference**: Optimized kernels, efficient scheduler loop, and Cache-DiT acceleration
- **Ease of Use**: OpenAI-compatible API, CLI, and Python SDK
- **Multi-Platform**: NVIDIA GPUs (H100, H200, A100, B200, 4090) and AMD GPUs (MI300X, MI325X)
- **Multi-Platform**: NVIDIA GPUs (H100, H200, A100, B200, 4090), AMD GPUs (MI300X, MI325X) and Ascend NPU (A2, A3)
---

4
docs/diffusion/installation.md
View File

@@ -82,6 +82,10 @@ rm -f python/pyproject.toml && mv python/pyproject_other.toml python/pyproject.t
pip install -e "python[all_musa]"
```
## Platform-Specific: Ascend NPU
For Ascend NPU, please follow the [NPU installation guide](../platforms/ascend_npu.md).
Quick test:
```bash

23
docs/diffusion/performance/attention_backends.md
View File

@@ -14,6 +14,7 @@ When using the diffusers backend, `--attention-backend` is passed through to dif
- **CUDA**: prefers FlashAttention (FA3/FA4) when supported; otherwise falls back to PyTorch SDPA.
- **ROCm**: uses FlashAttention when available; otherwise falls back to PyTorch SDPA.
- **MPS**: always uses PyTorch SDPA.
- **NPU**: always uses PyTorch SDPA.
## Backend options
@@ -83,17 +84,17 @@ Some backends require additional configuration. You can pass these parameters vi
## Platform support matrix
| Backend | CUDA | ROCm | MPS | Notes |
|---|---:|---:|---:|---|
| `fa` | ✅ | ✅ | ❌ | CUDA requires SM80+ and fp16/bf16. FlashAttention is only used when the required runtime is installed; otherwise it falls back to `torch_sdpa`. |
| `torch_sdpa` | ✅ | ✅ | ✅ | Most compatible option across platforms. |
| `sliding_tile_attn` | ✅ | ❌ | ❌ | CUDA-only. Requires `st_attn`. Configure via `--attention-backend-config`. |
| `sage_attn` | ✅ | ❌ | ❌ | CUDA-only (optional dependency). |
| `sage_attn_3` | ✅ | ❌ | ❌ | CUDA-only (optional dependency). |
| `video_sparse_attn` | ✅ | ❌ | ❌ | CUDA-only. Requires `vsa`. Configure `sparsity` via `--attention-backend-config`. |
| `vmoba_attn` | ✅ | ❌ | ❌ | CUDA-only. Requires `kernel.attn.vmoba_attn.vmoba`. Configure via `--attention-backend-config`. |
| `aiter` | ✅ | ❌ | ❌ | Requires `aiter`. |
| `sparse_video_gen_2_attn` | ✅ | ❌ | ❌ | CUDA-only. Requires `svg`. |
| Backend | CUDA | ROCm | MPS | NPU | Notes |
|---|---:|---:|---:|---:|---|
| `fa` | ✅ | ✅ | ❌ | ❌ | CUDA requires SM80+ and fp16/bf16. FlashAttention is only used when the required runtime is installed; otherwise it falls back to `torch_sdpa`. |
| `torch_sdpa` | ✅ | ✅ | ✅ | ✅ | Most compatible option across platforms. |
| `sliding_tile_attn` | ✅ | ❌ | ❌ | ❌ | CUDA-only. Requires `st_attn`. Configure via `--attention-backend-config`. |
| `sage_attn` | ✅ | ❌ | ❌ | ❌ | CUDA-only (optional dependency). |
| `sage_attn_3` | ✅ | ❌ | ❌ | ❌ | CUDA-only (optional dependency). |
| `video_sparse_attn` | ✅ | ❌ | ❌ | ❌ | CUDA-only. Requires `vsa`. Configure `sparsity` via `--attention-backend-config`. |
| `vmoba_attn` | ✅ | ❌ | ❌ | ❌ | CUDA-only. Requires `kernel.attn.vmoba_attn.vmoba`. Configure via `--attention-backend-config`. |
| `aiter` | ✅ | ❌ | ❌ | ❌ | Requires `aiter`. |
| `sparse_video_gen_2_attn` | ✅ | ❌ | ❌ | ❌ | CUDA-only. Requires `svg`. |
## Usage