From 30191ed112d457fdabbe9ab648c1bb45838b525a Mon Sep 17 00:00:00 2001
From: Didan Deng <33117903+wtomin@users.noreply.github.com>
Date: Sat, 7 Mar 2026 01:45:02 +0800
Subject: [PATCH] [Feature]:  Add vae-patch-parallel CLI argument in online
 serving (#1716)

Signed-off-by: Didan Deng <33117903+wtomin@users.noreply.github.com>
---
 .../examples/online_serving/text_to_image.md  | 23 +++++++++++++++++++
 vllm_omni/entrypoints/async_omni.py           |  2 ++
 vllm_omni/entrypoints/cli/serve.py            |  9 ++++++++
 3 files changed, 34 insertions(+)

diff --git a/docs/user_guide/examples/online_serving/text_to_image.md b/docs/user_guide/examples/online_serving/text_to_image.md
index be8c7d894..793129488 100644
--- a/docs/user_guide/examples/online_serving/text_to_image.md
+++ b/docs/user_guide/examples/online_serving/text_to_image.md
@@ -23,6 +23,29 @@ Or use the startup script:
 bash run_server.sh
 ```
 
+### Start with Parallelism Acceleration
+
+Enable Tensor Parallelism and VAE Patch Parallelism for faster inference:
+
+```bash
+# With Tensor Parallelism (requires >= 2 GPUs)
+vllm serve Qwen/Qwen-Image --omni --port 8091 --tensor-parallel-size 2
+
+# With Tensor Parallelism and VAE Patch Parallelism (requires >= 2 GPUs)
+vllm serve Qwen/Qwen-Image --omni --port 8091 --tensor-parallel-size 2 --vae-patch-parallel-size 2 --vae-use-tiling
+
+# With Sequence Parallelism (Ulysses-SP, requires >= 2 GPUs)
+vllm serve Qwen/Qwen-Image --omni --port 8091 --usp 2
+
+# With Ring-Attention (requires >= 2 GPUs)
+vllm serve Qwen/Qwen-Image --omni --port 8091 --ring 2
+
+# Combined: Ulysses + Ring (requires >= 4 GPUs)
+vllm serve Qwen/Qwen-Image --omni --port 8091 --usp 2 --ring 2
+```
+
+For more details on parallelism acceleration, see the [Parallelism Acceleration Guide](../../diffusion/parallelism_acceleration.md).
+
 ## API Calls
 
 ### Method 1: Using curl
diff --git a/vllm_omni/entrypoints/async_omni.py b/vllm_omni/entrypoints/async_omni.py
index 6939987b0..e393f31be 100644
--- a/vllm_omni/entrypoints/async_omni.py
+++ b/vllm_omni/entrypoints/async_omni.py
@@ -141,6 +141,7 @@ class AsyncOmni(OmniBase):
             sequence_parallel_size = kwargs.get("sequence_parallel_size")
             tensor_parallel_size = kwargs.get("tensor_parallel_size") or 1
             cfg_parallel_size = kwargs.get("cfg_parallel_size") or 1
+            vae_patch_parallel_size = kwargs.get("vae_patch_parallel_size") or 1
             use_hsdp = kwargs.get("use_hsdp", False)
             hsdp_shard_size = kwargs.get("hsdp_shard_size", -1)
             hsdp_replicate_size = kwargs.get("hsdp_replicate_size", 1)
@@ -165,6 +166,7 @@ class AsyncOmni(OmniBase):
                 ulysses_degree=ulysses_degree,
                 ring_degree=ring_degree,
                 cfg_parallel_size=cfg_parallel_size,
+                vae_patch_parallel_size=vae_patch_parallel_size,
                 use_hsdp=use_hsdp,
                 hsdp_shard_size=hsdp_shard_size,
                 hsdp_replicate_size=hsdp_replicate_size,
diff --git a/vllm_omni/entrypoints/cli/serve.py b/vllm_omni/entrypoints/cli/serve.py
index a580f9536..7365e73bb 100644
--- a/vllm_omni/entrypoints/cli/serve.py
+++ b/vllm_omni/entrypoints/cli/serve.py
@@ -219,6 +219,7 @@ class OmniServeCommand(CLISubcommand):
         )
         omni_config_group.add_argument(
             "--ring",
+            "--ring-degree",
             dest="ring_degree",
             type=int,
             default=None,
@@ -335,6 +336,14 @@ class OmniServeCommand(CLISubcommand):
             help="Number of devices for CFG parallel computation for diffusion models. "
             "Equivalent to setting DiffusionParallelConfig.cfg_parallel_size.",
         )
+        omni_config_group.add_argument(
+            "--vae-patch-parallel-size",
+            type=int,
+            default=1,
+            help="VAE Patch Parallelism degree for diffusion models. "
+            "Distributes VAE decode workload across multiple ranks by splitting the latent spatially. "
+            "Equivalent to setting DiffusionParallelConfig.vae_patch_parallel_size.",
+        )
 
         # Default sampling parameters
         omni_config_group.add_argument(