diff --git a/paddle/cuda/include/hl_matrix_type.cuh b/paddle/cuda/include/hl_matrix_type.cuh index 2ced2fb1ab1afc8b0619a904709ce4504291dd86..77f73167fe6ea2e301b3f68e51d32f3e1578ec93 100644 --- a/paddle/cuda/include/hl_matrix_type.cuh +++ b/paddle/cuda/include/hl_matrix_type.cuh @@ -38,7 +38,7 @@ typedef double2 vecType; #endif #elif defined(__SSE3__) #include "hl_cpu_simd_sse.cuh" -#elif defined(__ARM_NEON) || defined(__ARM_NEON__) +#elif (defined(__ARM_NEON) || defined(__ARM_NEON__)) && !defined(__NVCC__) #include "hl_cpu_simd_neon.cuh" #else #include "hl_cpu_scalar.cuh"