@
ltltfuture 好希望早日支持 pytorch,不过 pytorch 似乎也已经在用 Metal Performance Shaders 加速训练和运行了(
https://github.com/pytorch/pytorch/pull/46112 )...
@
agagega 目前 ML Compute 只支持 CPU 和 GPU,但是 ML Compute 里面有 ANE 相关的 symbol,所以不排除是目前 Neural Engine 的加速不成熟(因为 MLC 是 macOS 11 和 iOS 14 刚加进来的),以后可能就会支持...
```
00007fff2a15aae2 t +[MLCDeviceANE hasANE]
00007fff2a22b934 t +[MLCDeviceANE hasANE].cold.1
00007fff2a15b672 t -[MLCDeviceANE .cxx_destruct]
00007fff2a15ad57 t -[MLCDeviceANE allocateDeviceMemoryForTensor:]
00007fff2a22bcad t -[MLCDeviceANE allocateDeviceMemoryForTensor:].cold.1
00007fff2a15b05c t -[MLCDeviceANE broadcastTensor:]
00007fff2a22c124 t -[MLCDeviceANE broadcastTensor:].cold.1
00007fff2a15b0b1 t -[MLCDeviceANE broadcastTensor:batchSize:]
00007fff2a22c1a3 t -[MLCDeviceANE broadcastTensor:batchSize:].cold.1
00007fff2a15b309 t -[MLCDeviceANE commitAndWaitForCompletion:]
00007fff2a22c51c t -[MLCDeviceANE commitAndWaitForCompletion:].cold.1
00007fff2a15b35e t -[MLCDeviceANE commitAndWaitForCompletion:enableProfiling:graphExecutionTime:graphResultTensor:]
00007fff2a22c59b t -[MLCDeviceANE commitAndWaitForCompletion:enableProfiling:graphExecutionTime:graphResultTensor:].cold.1
...
```