[diffusion] docs: sync snapshot and minimax-h3 subblock features (#38784)

Co-authored-by: Mick Qian <mickqian@users.noreply.github.com>

docs/cookbook/diffusion/MiniMax/MiniMax-H3.mdx

@@ -1187,6 +1187,59 @@ benchmark latency as well as visual and audio quality on the target workload.
 
 </Tab>
 
+<Tab title="SubBlock sparse attention">
+
+SubBlock sparse attention applies training-free block sparsity to MiniMax-H3's
+long, non-causal DiT self-attention. The BF16 path supports CUDA SM90, SM100,
+and SM120. Other compute capabilities, including SM103/B300, are rejected.
+Ulysses sequence parallelism is supported; Ring parallelism is not.
+
+The global backend selects the H3 DiT lazily, so keep the Qwen text encoder on
+its compatible dense backend with a component override:
+
+```bash Command
+sglang serve \
+  --model-path MiniMaxAI/MiniMax-H3 \
+  --model-variant fl2va \
+  --num-gpus 8 \
+  --ulysses-degree 8 \
+  --attention-backend subblock_sparse_attn \
+  --component-attention-backends text_encoder=fa \
+  --attention-backend-config '{"sparsity": 0.75, "skip_first_steps": 10}' \
+  --port 30010
+```
+
+Use `text_encoder=torch_sdpa` on SM120. The default configuration drops at
+most 75% of key blocks after the first 10 denoise forwards; short sequences,
+cross-attention, and unsupported shapes continue on the dense path. Tune
+`sparsity` and `skip_first_steps` together with visual and audio quality checks.
+
+On H100/H200 (SM90), `compute_mode=sage_fp8` switches the sparse kernel to
+online INT8 Q/K and FP8 P/V compute. Install the optional kernel and add the
+compute mode:
+
+```bash Command
+pip install git+https://github.com/thu-ml/SpargeAttn.git --no-build-isolation
+
+sglang serve \
+  --model-path MiniMaxAI/MiniMax-H3 \
+  --model-variant fl2va \
+  --num-gpus 8 \
+  --ulysses-degree 8 \
+  --attention-backend subblock_sparse_attn \
+  --component-attention-backends text_encoder=fa \
+  --attention-backend-config '{"compute_mode": "sage_fp8", "sparsity": 0.75}' \
+  --port 30010
+```
+
+<Warning>
+Both SubBlock routing and `sage_fp8` are approximate. Use dense BF16 attention
+for consistency ground truth, and validate the selected sparsity on the target
+resolution, duration, task, and checkpoint.
+</Warning>
+
+</Tab>
+
 <Tab title="Encoder scheduling">
 
 The picker explicitly writes `--encoder-parallel auto` in every single-node

docs/docs/sglang-diffusion/attention_backends.mdx

@@ -79,6 +79,11 @@ For SGLang-native pipelines, the CLI accepts the lowercase names of `AttentionBa
       <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)", whiteSpace: "nowrap"}}>`VIDEO_SPARSE_ATTN`</td>
       <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Requires <code>vsa</code>. Configure <code>sparsity</code> via <code>--attention-backend-config</code>.</td>
     </tr>
+    <tr>
+      <td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>`subblock_sparse_attn`</td>
+      <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)", whiteSpace: "nowrap"}}>`SUBBLOCK_SPARSE_ATTN`</td>
+      <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Training-free SubBlock sparsity for MiniMax-H3 on CUDA SM90, SM100, and SM120. The default compute mode is BF16; <code>compute_mode=sage_fp8</code> is an approximate SM90-only path that requires <code>SpargeAttn</code>. See the <a href="/cookbook/diffusion/MiniMax/MiniMax-H3#subblock-sparse-attention">MiniMax-H3 recipe</a>.</td>
+    </tr>
     <tr>
       <td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>`video_sparse_attn_h3`</td>
       <td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)", whiteSpace: "nowrap"}}>`VIDEO_SPARSE_ATTN_H3`</td>

python/sglang/multimodal_gen/README.md

@@ -84,6 +84,8 @@ loaded component:
 
 - `resident` keeps the complete component on the accelerator.
 - `component-offload` stores the complete component on CPU between uses.
+- `snapshot-offload` keeps a CPU weight snapshot while the complete component
+  runs on the accelerator, avoiding a weight copy back to CPU after each use.
 - `layerwise-offload` streams the component's declared layers from CPU.
 
 `COMPONENT` can be an exact `model_index.json` key or one of `all`, `dit`,