OpenHarmony/third_party_mesa3d
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

iris: Don't allocate a BO per query object

Browse Source 
Instead of allocating 4K BO per query object, we can create a large blob
of memory and split it into pieces as required.

Having one BO for multiple query objects, we don't want to wait on all
of them, instead when we write last snapshot, we create a sync point, and
check syncpoints while waiting on particular object.

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
This commit is contained in:
Sagar Ghuge
2019-01-15 14:15:07 -08:00
committed by Kenneth Graunke
parent a1ebac3750
commit c24a574e6c
5 changed files with 97 additions and 45 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
src/gallium/drivers/iris/iris_context.c
View File

@@ -113,6 +113,7 @@ iris_destroy_context(struct pipe_context *ctx)
iris_destroy_border_color_pool(ice);
u_upload_destroy(ice->state.surface_uploader);
u_upload_destroy(ice->state.dynamic_uploader);
u_upload_destroy(ice->query_buffer_uploader);
slab_destroy_child(&ice->transfer_pool);
@@ -195,6 +196,10 @@ iris_create_context(struct pipe_screen *pscreen, void *priv, unsigned flags)
u_upload_create(ctx, 16384, PIPE_BIND_CUSTOM, PIPE_USAGE_IMMUTABLE,
IRIS_RESOURCE_FLAG_DYNAMIC_MEMZONE);
ice->query_buffer_uploader =
u_upload_create(ctx, 4096, PIPE_BIND_CUSTOM, PIPE_USAGE_IMMUTABLE,
0);
genX_call(devinfo, init_state, ice);
genX_call(devinfo, init_blorp, ice);