diff options
| author | Zhenyu Wang <zhenyuw@linux.intel.com> | 2014-10-23 15:19:22 +0800 |
|---|---|---|
| committer | Zhigang Gong <zhigang.gong@intel.com> | 2014-10-24 18:12:29 +0800 |
| commit | 8b0850587b80847056c9d798642fd8a7155fe3c3 (patch) | |
| tree | e2eaae3f53d84ac8d344785290f49f55ee0372d6 /src/cl_command_queue.c | |
| parent | e74164f951ba9003f92d14d685f184617da056eb (diff) | |
Make use of write enable flag for mem bo map
Use drm/intel optimization for mem bo mapping in case of read or write.
So we could be possibly waiting less.
This also adds 'map_flags' check in clEnqueueMapBuffer/clEnqueueMapImage
for actual read or write mapping.
But currently leave clMapBufferIntel untouched which might break ABI/API.
v2: Fix write_map flag in clEnqueueMapBuffer/clEnqueueMapImage.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Reviewed-by: "Guo, Yejun" <yejun.guo@intel.com>
Reviewed-by: Zhigang Gong <zhigang.gong@linux.intel.com>
Diffstat (limited to 'src/cl_command_queue.c')
| -rw-r--r-- | src/cl_command_queue.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/cl_command_queue.c b/src/cl_command_queue.c index 48deba05..d07774f0 100644 --- a/src/cl_command_queue.c +++ b/src/cl_command_queue.c @@ -336,7 +336,7 @@ cl_fulsim_read_all_surfaces(cl_command_queue queue, cl_kernel k) assert(mem->bo); chunk_n = cl_buffer_get_size(mem->bo) / chunk_sz; chunk_remainder = cl_buffer_get_size(mem->bo) % chunk_sz; - to = cl_mem_map(mem); + to = cl_mem_map(mem, 1); for (j = 0; j < chunk_n; ++j) { char name[256]; sprintf(name, "dump%03i.bmp", curr); |