add dispatch_gmm_combine kernel (#3532)

### What this PR does / why we need it?

This PR introduces the Ascend implementation of the
`dispatch_ffn_combine` kernel and wires it into the vLLM-Ascend runtime,
together with follow‑up fixes to ensure the kernel builds and runs
correctly in CI.

- Add full host and device implementation of the `dispatch_ffn_combine`
kernel under `csrc/dispatch_ffn_combine`, including tiling logic, MOE
routing helpers, and kernel utilities for quantized FFN dispatch.
- Integrate the new kernel with the PyTorch binding
(csrc/torch_binding.cpp, csrc/torch_binding_meta.cpp) and the Ascend
runtime (vllm_ascend/ascend_forward_context.py,
vllm_ascend/worker/model_runner_v1.py).
- Extend fused MoE communication and token dispatch support in
`vllm_ascend/ops/fused_moe`, adding methods/utilities needed by the new
dispatch path.
- Update quantization logic in vllm_ascend/quantization/w8a8_dynamic.py
to support the new FFN dispatch flow.
- Fix kernel build issues by adjusting `csrc/build_aclnn.sh`, CMake
configuration, and include/namespace usage in the new kernel files.
- Add an end‑to‑end nightly test
`tests/e2e/nightly/ops/test_dispatch_ffn_combine.py` and helper
utilities in `vllm_ascend/utils.py` to validate the new kernel.

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?


- vLLM version: v0.12.0
- vLLM main: https://github.com/vllm-project/vllm/commit/v0.12.0

---------

Signed-off-by: mojave2 <chenchen145@huawei.com>
Co-authored-by: wangxiyuan <wangxiyuan1007@gmail.com>

vllm_ascend/utils.py

@@ -911,6 +911,9 @@ def get_hccl_config_for_pg_options(group_name: str) -> Optional[dict]:
         "dp": {
             "hccl_buffer_size": calculate_dp_buffer_size()
         },
+        "ep": {
+            "hccl_buffer_size": calculate_ep_buffer_size()
+        },
     }
     return hccl_config_map.get(group_name, get_default_buffer_config())
 
@@ -932,6 +935,30 @@ def calculate_dp_buffer_size() -> int:
     return max(dp_buffer_size, _MIN_DP_BUFFER_SIZE)
 
 
+def calculate_ep_buffer_size() -> int:
+    """
+    formula of ep buffer size:
+    batch_size * hidden_size * topk * 4
+    """
+    ep_buffer_size = _DEFAULT_BUFFER_SIZE
+    try:
+        from vllm.config import get_current_vllm_config
+        vllm_config = get_current_vllm_config()
+        hf_config = vllm_config.model_config.hf_config
+
+        hidden_size = hf_config.hidden_size
+        topk = getattr(hf_config, "num_experts_per_token", 1)
+        batch_size = vllm_config.scheduler_config.max_num_batched_tokens
+        int8_size = torch.iinfo(torch.int8).bits // 8
+        bf16_size = torch.finfo(torch.bfloat16).bits // 8
+        ep_buffer_size = math.ceil(
+            (batch_size * hidden_size * topk *
+             (int8_size * 2 + bf16_size)) / (1024 * 1024))
+    except Exception:
+        pass
+    return max(ep_buffer_size, _DEFAULT_BUFFER_SIZE)
+
+
 # Currently, when in A2, setting the environment variables HCCL_INTRA_PCIE_ENABLE=1
 # and HCCL_INTRA_ROCE_ENABLE=0 can reduce cross-machine communication traffic and
 # significantly improve communication performance of MC2 ops dispatch/combine.