Set EOP buffer TC policy to non-coherent

Restores regression in dispatch latency.

Change-Id: I17869d3d515d8c1fa055a57afec2531903b88b16
Signed-off-by: Jay Cornwall <Jay.Cornwall@amd.com>


[ROCm/ROCR-Runtime commit: b764991982]

projects/rocr-runtime/src/queues.c

@@ -388,6 +388,7 @@ void *allocate_exec_aligned_memory_gpu(uint32_t size, uint32_t align,
 	flags.ui32.ExecuteAccess = 1;
 	flags.ui32.NonPaged = nonPaged;
 	flags.ui32.PageSize = HSA_PAGE_SIZE_4KB;
+	flags.ui32.CoarseGrain = DeviceLocal;
 
 	size = ALIGN_UP(size, align);