华为昇腾310p编译安装出现CANN版本不匹配的问题

环境

• 【FastDeploy版本】： 编译安装develop分支
• 【编译命令】cmake -DWITH_ASCEND=ON -DCMAKE_INSTALL_PREFIX=fastdeploy-ascend -DENABLE_VISION=ON ..
• 【系统平台】: Linux worker-12 4.19.90-23.8.v2101.ky10.x86_64
• 【硬件】： Ascend 310I Pro（310P3）
• 【编译语言】：Python 3.10
• 【CANN版本】：CANN 6.4.12.1.241:6.3.RC2
• 【npu 驱动版本】：Version: 23.0.rc2

问题日志及出现问题的操作流程

• 【模型跑不通】
  -- 首先编译安装Paddle-Lite，编译命令：$./lite/tools/build_linux.sh --arch=x86 --with_extra=ON --with_log=ON --with_exception=ON --with_nnadapter=ON --nnadapter_with_huawei_ascend_npu=ON --nnadapter_huawei_ascend_npu_sdk_root=/usr/local/Ascend/ascend-toolkit/latest full_publish
  -- 修改FastDeploy/cmake/paddlelite.cmake中的内容：set(PADDLELITE_DIRECTORY /home/xxx/Paddle-Lite/build.lite.linux.x86.gcc/inference_lite_lib/cxx CACHE PATH "Directory of custom Paddle-Lite library")设置为编译好的Paddle-Lite库
  -- 根据示例进行C++和python编译安装Fastdeploy（https://github.com/PaddlePaddle/FastDeploy/blob/develop/docs/cn/build_and_install/huawei_ascend.md）
  -- 运行example示例（https://github.com/PaddlePaddle/FastDeploy/blob/develop/examples/vision/detection/paddledetection/python/README_CN.md）
  -- 执行命令：

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/xxx/Paddle-Lite/build.lite.linux.x86.gcc/inference_lite_lib/third_party/mklml/lib/
python infer_ppyoloe.py --model_dir ppyoloe_crn_l_300e_coco --image 000000014439.jpg --device ascend

无法完成昇腾芯片中的推理，错误信息如下：
I1108 13:48:41.355319 730851 generate_program_pass.h:41] insts.size: 1
[INFO] fastdeploy/runtime/runtime.cc(354)::CreateLiteBackend Runtime initialized with Backend::PDLITE in Device::ASCEND.
[W 11/ 8 13:48:41.490 .../src/driver/huawei_ascend_npu/utility.cc:57 InitializeAscendCL] CANN version mismatch. The build version is 0.0.0, but the current environment version is 6.3.2.
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:42 Context] properties:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:67 Context] selected device ids:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:69 Context] 0
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:79 Context] profiling path:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:89 Context] dump model path:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:99 Context] precision mode:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:121 Context] op select impl mode:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:131 Context] op type list for impl mode:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:141 Context] enable compressw weight:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:151 Context] auto tune mode:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:161 Context] enable dynamic shape range:
[I 11/ 8 13:48:41.677 ...r/src/driver/huawei_ascend_npu/engine.cc:177 Context] initial buffer length of dynamic shape range: -1
[W 11/ 8 13:48:41.678 ...ter/nnadapter/src/runtime/compilation.cc:334 Finish] Warning: Failed to create a program, No model and cache is provided.
W1108 13:48:41.678025 730851 engine.cc:149] Warning: Build model failed(3) !
[W 11/ 8 13:48:41.693 ...nnadapter/nnadapter/src/runtime/model.cc:86 GetSupportedOperations] Warning: Failed to get the supported operations for device 'huawei_ascend_npu', because the HAL interface 'validate_program' is not implemented!
W1108 13:48:41.693092 730851 converter.cc:171] Warning: Failed to get the supported operations for the selected devices, one or more of the selected devices are not supported!
[I 11/ 8 13:48:41.693 ...r/src/driver/huawei_ascend_npu/driver.cc:70 CreateProgram] Create program for huawei_ascend_npu.
[F 11/ 8 13:48:42.749 .../src/driver/huawei_ascend_npu/utility.cc:315 BuildOMModelToBuffer] Check failed: (reinterpret_castge::graphStatus(aclgrphBuildModel(ir_graph, options, om_buffer)) == ge::GRAPH_SUCCESS): 1343266818!==0 1343266818 Unknown ATC error code(1343266818)
[F 11/ 8 13:48:42.749 .../src/driver/huawei_ascend_npu/utility.cc:315 BuildOMModelToBuffer] Check failed: (reinterpret_castge::graphStatus(aclgrphBuildModel(ir_graph, options, om_buffer)) == ge::GRAPH_SUCCESS): 1343266818!==0 1343266818 Unknown ATC error code(1343266818)

[F 11/ 8 13:48:42.854 ...ter/nnadapter/src/runtime/compilation.cc:98 ~Program] Check failed: device_context: No device found.
[F 11/ 8 13:48:42.854 ...ter/nnadapter/src/runtime/compilation.cc:98 ~Program] Check failed: device_context: No device found.

已放弃 (核心已转储)

Start with cmake/paddlelite.cmake, the Huawei Ascend build guide, and the referenced detection example README. Reproduce the C++ and Python build and inference commands, then inspect the reported version check in utility.cc alongside the engine and driver logs. Done means the example runs on Ascend 310P without the CANN mismatch and ATC failures.