flywithu/ntel-gpu-tools
1
0
Fork 0
mirror of https://github.com/tiagovignatti/intel-gpu-tools.git synced 2025-11-03 19:47:15 +00:00
Code Issues Releases Wiki Activity

lib: Apply magic clflush serialisation

Browse Source 
On Baytrail, Braswell and Atoms beyond we see an issue where the mfence
is insufficient to force the cacheline to be coherent (i.e. such that
writes from the GPU are visible by the CPU after the call to clflush). A
second clflush is ordered with an earlier clflush to the same address
and this appears sufficient to give the coherency required for GPU/CPU
interop.

Testcase: igt/gem_exec_flush
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Akash Goel <akash.goel@intel.com>
This commit is contained in:
Chris Wilson 2016-05-01 09:07:29 +01:00
parent ef3f61a6d3
commit b481705208
1 changed files with 1 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
lib/igt_gt.c
View File

@ -425,6 +425,7 @@ void igt_clflush_range(void *addr, int size)
__builtin_ia32_mfence(); __builtin_ia32_mfence();
for (; p < end; p += clflush_size) for (; p < end; p += clflush_size)
__builtin_ia32_clflush(p); __builtin_ia32_clflush(p);
__builtin_ia32_clflush(end - 1); /* magic serialisation for byt+ */
__builtin_ia32_mfence(); __builtin_ia32_mfence();
#else #else
fprintf(stderr, "igt_clflush_range() unsupported\n"); fprintf(stderr, "igt_clflush_range() unsupported\n");