参考代码：https://gitcode.com/cann/asc-devkit/blob/master/examples/01_simd_cpp_api/00_introduction/02_matrix/matmul_advanced_api/README.md
原示例标注支持的产品不包含 310P3 (Atlas 推理系列产品），无法直接搬运使用
根据文档 Matmul 算子是支持推理卡int8 矩阵相乘的： https://www.hiascend.com/document/detail/zh/canncommercial/latest/API/ascendcopapi/atlasascendc_api_07_0614.html
我的测试代码是根据文档写的，代码使用接口也符合规范，但是报错
参考代码：https://gitcode.com/cann/asc-devkit/blob/master/examples/01_simd_cpp_api/00_introduction/02_matrix/matmul_advanced_api/README.md
原示例标注支持的产品不包含 310P3 (Atlas 推理系列产品），无法直接搬运使用
参考代码：https://gitcode.com/cann/asc-devkit/blob/master/examples/01_simd_cpp_api/00_introduction/02_matrix/matmul_advanced_api/README.md
原示例标注支持的产品不包含 310P3 (Atlas 推理系列产品），无法直接搬运使用
多谢，310P3需要用旧的示例能跑通：https://gitee.com/ascend/samples/tree/master/operator/ascendc/tutorials/MatmulCustomSample/KernelLaunch
Thanks for sending an issue! Please fill in the following template to help quickly solve your problem.
Describe the current behavior / 问题描述 (Mandatory / 必填)
参考矩阵乘法实例出错：
参考代码：https://gitcode.com/cann/asc-devkit/blob/master/examples/01_simd_cpp_api/00_introduction/02_matrix/matmul_advanced_api/README.md
原示例是半浮点相乘，我把输入矩阵换成int8时就报错，运行硬件是310P3
代码及日志见gitee：https://gitee.com/luyaoo/ascend_matmul
运行出错日志:
matmul_int8/matmul_advanced_api.asc FUNC: aclrtSynchronizeStream(stream) LINE: 157, acl error, errorCode = 507015, EZ9999: Inner Error!
VEC instruction error: the destination blocks have the same address.[FUNC:PrintCoreErrorInfo][FILE:device_error_core_proc.cc][LINE:227]
DEVICE[0] PID[3809974]:
EXCEPTION STREAM:
Exception info:TGID=3809974, model id=65535, stream id=20, stream phase=SCHEDULE
Environment / 环境信息 (Mandatory / 必填)
+--------------------------------------------------------------------------------------------------------+
| npu-smi 26.0.rc1 Version: 26.0.rc1 |
+-------------------------------+-----------------+------------------------------------------------------+
| NPU Name | Health | Power(W) Temp(C) Hugepages-Usage(page) |
| Chip Device | Bus-Id | AICore(%) Memory-Usage(MB) |
+===============================+=================+======================================================+
| 3 310P3 | OK | NA 45 0 / 0 |
| 0 12 | 0000:2F:00.0 | 0 1344 / 44278 |
+-------------------------------+-----------------+------------------------------------------------------+
| 3 310P3 | OK | NA 44 0 / 0 |
| 1 13 | 0000:2F:00.0 | 0 1616 / 43693 |
+===============================+=================+======================================================+
ii ascend-cann-310p-ops 9.1.0-beta.3 arm64 ascend-cann-310p-ops
ii ascend-cann-nnal 9.1.0-beta.3 arm64 ascend-cann-nnal
ii ascend-cann-toolkit 9.1.0-beta.3 arm64 ascend-cann-toolkit
ii ascend310p-driver 26.0.rc1 arm64 Huawei Ascend310P Driver
Steps to reproduce the issue / 重现步骤 (Mandatory / 必填)
代码及日志见gitee：https://gitee.com/luyaoo/ascend_matmul
日志在log目录下
Describe the expected behavior / 预期结果 (Mandatory / 必填)
得到正确的Int8矩阵乘法结果
Related log / screenshot / 日志 / 截图 (Mandatory / 必填)
Thanks for sending an issue! Please fill in the following template to help quickly solve your problem.
Special notes for this issue/备注 (Optional / 选填)
Thanks for sending an issue! Please fill in the following template to help quickly solve your problem.