SWDEV-252337: Fix for V100 compilation issue

Change-Id: Ib57ff3ac28f0a581f395f32181974e64eca6d858


[ROCm/hip commit: 0be1f7843a]

projects/hip/docs/markdown/hip_faq.md

@@ -247,4 +247,4 @@ The workaround is to explicitly add the keyword of "static" before any functions
 Product of block.x, block.y, and block.z should be less than 1024.
 
 ### Are __shfl_*_sync functions supported on HIP platform?
-__shfl_*_sync is not supported on HIP but for nvcc path CUDA 9.0 and above all shuffle calls get redirected to it's sync version.
+__shfl_*_sync is not supported on HIP but for nvcc path CUDA 9.0 and above all shuffle calls get redirected to it's sync version.

projects/hip/include/hip/nvcc_detail/hip_runtime_api.h

@@ -26,6 +26,7 @@ THE SOFTWARE.
 #include <cuda_runtime_api.h>
 #include <cuda.h>
 #include <cuda_profiler_api.h>
+#include <cuda_fp16.h>
 
 #ifdef __cplusplus
 extern "C" {