The source command on this line will only affect the environment of the script's execution shell. When this script is executed, it runs in a sub-shell, and any environment variables set within it are lost when the script finishes. If the intention is to modify the environment of the calling shell, this script should be sourced (e.g., source csrc/build_aclnn.sh) rather than executed. The #!/bin/bash shebang is misleading if the script is meant to be sourced. This can lead to silent failures in the environment setup.

The calculation of tensor strides is incorrect as it assumes the tensor is contiguous. This will lead to incorrect memory access and data corruption for non-contiguous tensors. You should use the tensor's actual strides and storage offset provided by PyTorch via at_tensor.strides() and at_tensor.storage_offset().

    const auto dimNum = at_tensor.dim();
    aclFormat format = ACL_FORMAT_ND;

    // 适配dispatch_gmm_combine_decode算子的weight入参
    if (acl_data_type == ACL_INT8 && dimNum == 3) {
        format = ACL_FORMAT_FRACTAL_NZ;
    }

    auto acl_tensor =
        aclCreateTensor(at_tensor.sizes().data(), dimNum, acl_data_type, at_tensor.strides().data(),
                        at_tensor.storage_offset(), format, at_tensor.sizes().data(), dimNum,
                        const_cast<void *>(at_tensor.storage().data()));

There appears to be a duplicated field aivNum and a confusing comment. The struct DispatchGmmCombineDecodeInfo has aicNum with comment // aivNum and aivNum with comment // aivNum. This is likely a copy-paste error and can lead to confusion and bugs. Please clarify the purpose of each field and correct the comments. For example, aicNum should probably be for AI Core count and aivNum for AI Vector count.

this file is for model only, no need to add it here. @Potabk

can add a new job for the ops at the end of https://github.com/vllm-project/vllm-ascend/blob/main/.github/workflows/vllm_ascend_test_nightly_a3.yaml , That's enough for your PR; I'll do a more detailed restructuring later.

  custom-ops-tests:
    name: test ops
    if: always() && (github.event_name == 'schedule' || github.event_name == 'workflow_dispatch')
    needs: multi-node-tests
    strategy:
      fail-fast: false
      matrix:
        test_config:
          - name: custom-op-dispatch_gmm_combine_decode
            os: linux-aarch64-a3-16
            tests: tests/e2e/nightly/multicard_ops/test_dispatch_gmm_combine_decode.py
    uses: ./.github/workflows/_e2e_nightly_single_node.yaml
    with:
      runner: ${{ matrix.test_config.os }}
      vllm: 0.12.0
      image: 'swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:nightly-a3'
      tests: ${{ matrix.test_config.tests }}
      name: ${{ matrix.test_config.name }}

Done