PaddleCustomDevice issues

[MLU] cherry-pick range and flash_attn optimize to develop

1

RT

ShawnNew

contributor

[NPU] Add FP32 cases for big shape ut

1

[NPU] Add FP32 cases for big shape ut

zlzzl

contributor

[DONT MERGE][MLU] tmp fix memcpy kernel.

1

RT

ShawnNew

contributor

[MLU] optimize range kernel; flash_attn kernel

1

1. make range kernel a function, which uses cnnlArange_v2 2. use range function in flash_attn

ShawnNew

contributor

华为910A 编译报错：include "unsupported/Eigen/CXX11/Tensor

14

/opt/py39/lib/python3.9/site-packages/paddle/include/paddle/phi/kernels/funcs/eigen/extensions.h:23:10: fatal error: unsupported/Eigen/CXX11/Tensor: No such file or directory #include "unsupported/Eigen/CXX11/Tensor" ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ compilation terminated. CMakeFiles/paddle-custom-npu.dir/build.make:257: recipe for target 'CMakeFiles/paddle-custom-npu.dir/kernels/argsort_kernel.cc.o' failed make[2]: *** [CMakeFiles/paddle-custom-npu.dir/kernels/argsort_kernel.cc.o] Error 1 CMakeFiles/paddle-custom-npu.dir/build.make:453: recipe for target 'CMakeFiles/paddle-custom-npu.dir/kernels/contiguous_kernel.cc.o'...

Jakin-huang