华为昇腾310p3上 subgraph算子不适配

[Dzhhdz]

完全按照官方文档进行docker build 和docker run

源码编译v2.12，编译指令为：
build_linux.sh --arch=armv8 --toolchain=gcc --with_extra=ON --with_python=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/5.1.RC1.alpha001 --nnadapter_huawei_ascend_npu_sdk_version=5.1.RC1.alpha001 full_publish

编译完成后安装了python 3.10
安装了paddlepaddle x2paddle
将onnx模型转成nb

运行程序报错如下：
--- Testing: AgAACg4XJf47Mm8aSKNCuYpV8ftzoMf8.png --- [1/41]
[W 7/ 2 1:29: 3.682 .../src/driver/huawei_ascend_npu/utility.cc:52 InitializeAscendCL] CANN version mismatch. The build version is 0.0.0, but the current environment version is 5.1.1.
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:41 Context] properties: HUAWEI_ASCEND_NPU_SELECTED_DEVICE_IDS=0
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:66 Context] selected device ids:
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:68 Context] 0
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:78 Context] profiling path:
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:88 Context] dump model path:
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:98 Context] precision mode:
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:120 Context] op select impl mode:
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:130 Context] op type list for impl mode:
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:140 Context] enable compressw weight:
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:150 Context] auto tune mode:
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:160 Context] enable dynamic shape range:
[I 7/ 2 1:29: 3.719 ...r/src/driver/huawei_ascend_npu/engine.cc:176 Context] initial buffer length of dynamic shape range: -1
[W 7/ 2 1:29: 3.719 ...ter/nnadapter/src/runtime/compilation.cc:334 Finish] Warning: Failed to create a program, No model and cache is provided.
[W 7/ 2 1:29: 3.719 ...le-Lite/lite/kernels/nnadapter/engine.cc:149 LoadFromCache] Warning: Build model failed(3) !
[F 7/ 2 1:29: 3.719 ...kernels/nnadapter/converter/converter.cc:109 Apply] Op converter 'subgraph' not found!
[F 7/ 2 1:29: 3.719 ...kernels/nnadapter/converter/converter.cc:109 Apply] Op converter 'subgraph' not found!

处理图片 AgAACg4XJf47Mm8aSKNCuYpV8ftzoMf8.png 时发生严重错误: Paddle-Lite C++ Exception:
[F 7/ 2 1:29: 3.719 ...kernels/nnadapter/converter/converter.cc:109 Apply] Op converter 'subgraph' not found!

算法模型是自己的，基于yolo8修改的，利用paddlelite的推理框架
我看官方文档好像华为昇腾上不支持subgraph算子，能不能尽快适配呢，谢谢
但是这个CANN版本mismatch我有点不理解

[hong19860320]

就是运行时的 CANN 版本和 paddlelite 编译的 CANN 版本不匹配，建议在我们提供的 docker 环境里运行：
https://www.paddlepaddle.org.cn/lite/v2.12/demo_guides/huawei_ascend_npu.html

或者需要将机器的已安装的 CANN 环境卸载，然后安装文档指定的 5.1.RC1.alpha001 CANN 包。