diff options
| author |  Rasmus Munk Larsen <rmlarsen@google.com> | 2019-06-20 16:18:37 -0700 |
|---|---|---|
| committer | Rasmus Munk Larsen <rmlarsen@google.com> | 2019-06-20 16:18:37 -0700 |
| commit | b8f8dac4ebddd2ca8db596fd7d67e4b35ee553be (patch) | |
| tree | 0e31c35fcd179524c17ed0965c9e9b60ce390b1b /Eigen/src/Core/arch/GPU | |
| parent | e5e63c2cad4b829c80319abadf9dfdfbd1627cc2 (diff) | |
Update comment as suggested by tra@google.com.
Diffstat (limited to 'Eigen/src/Core/arch/GPU')
| -rw-r--r-- | Eigen/src/Core/arch/GPU/Half.h | 6 | ||||
| -rw-r--r-- | Eigen/src/Core/arch/GPU/PacketMath.h | 6 |
2 files changed, 6 insertions, 6 deletions
diff --git a/Eigen/src/Core/arch/GPU/Half.h b/Eigen/src/Core/arch/GPU/Half.h index c510d3beb..ffd4789cb 100644 --- a/Eigen/src/Core/arch/GPU/Half.h +++ b/Eigen/src/Core/arch/GPU/Half.h @@ -313,9 +313,9 @@ EIGEN_STRONG_INLINE __device__ bool operator >= (const half& a, const half& b) { #endif -// The EIGEN_COMP_CLANG && !EIGEN_COMP_NVCC clause is necessary on Mac where we -// suspect that NVCC is doing device-side preprocessing with clang, but actually -// compiling with cicc, which is obviously not clang. +// We need to distinguish ‘clang as the CUDA compiler’ from ‘clang as the host compiler, +// invoked by NVCC’ (e.g. on MacOS). The former needs to see both host and device implementation +// of the functions, while the latter can only deal with one one of them. #if !defined(EIGEN_HAS_NATIVE_FP16) || (EIGEN_COMP_CLANG && !EIGEN_COMP_NVCC) // Emulate support for half floats #if EIGEN_COMP_CLANG && defined(EIGEN_CUDACC) diff --git a/Eigen/src/Core/arch/GPU/PacketMath.h b/Eigen/src/Core/arch/GPU/PacketMath.h index d09795163..afc086090 100644 --- a/Eigen/src/Core/arch/GPU/PacketMath.h +++ b/Eigen/src/Core/arch/GPU/PacketMath.h @@ -102,9 +102,9 @@ template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE double2 pset1<double2>(const do return make_double2(from, from); } -// The EIGEN_COMP_CLANG && !EIGEN_COMP_NVCC clause is necessary on Mac where we -// suspect that NVCC is doing device-side preprocessing with clang, but actually -// compiling with cicc, which is obviously not clang. +// We need to distinguish ‘clang as the CUDA compiler’ from ‘clang as the host compiler, +// invoked by NVCC’ (e.g. on MacOS). The former needs to see both host and device implementation +// of the functions, while the latter can only deal with one one of them. #if defined(EIGEN_CUDA_ARCH) || defined(EIGEN_HIP_DEVICE_COMPILE) || (defined(EIGEN_CUDACC) && EIGEN_COMP_CLANG && !EIGEN_COMP_NVCC) namespace { |