[PATCH] D24946: [CUDA] Added support for CUDA-8

[Justin Lebar via cfe-commits]

jlebar added inline comments.

================
Comment at: lib/Headers/__clang_cuda_runtime_wrapper.h:139
@@ -137,1 +138,3 @@
 
+// CUDA 8.0.41 relies on __USE_FAST_MATH__ and __CUDA_PREC_DIV's values
+// Previous versions used to check thether they are defined or not.
----------------
Nit, missing period.

================
Comment at: lib/Headers/__clang_cuda_runtime_wrapper.h:140
@@ +139,3 @@
+// CUDA 8.0.41 relies on __USE_FAST_MATH__ and __CUDA_PREC_DIV's values
+// Previous versions used to check thether they are defined or not.
+// CU_DEVICE_INVALID macro is only defined in 8.0.41, so we use it
----------------
typo

================
Comment at: lib/Headers/__clang_cuda_runtime_wrapper.h:156
@@ +155,3 @@
+#endif
+#endif
+
----------------
I don't understand what we are doing here...

We're saying, if __USE_FAST_MATH__ is defined, and if it's not equal to 0, then redefine it equal to 1?  Isn't that a compile error?


https://reviews.llvm.org/D24946