flywithu/ntel-gpu-tools
1
0
Fork 0
mirror of https://github.com/tiagovignatti/intel-gpu-tools.git synced 2025-06-25 00:36:16 +00:00
Code Issues Releases Wiki Activity

lib/igt_draw: break if we already wrote every pixel

Browse Source 
Due to the nature of accessing a tiled buffer in an untiled way, we
used to loop through the whole buffer all the time. Add a small
mechanism to just break in case we know we already wrote every pixel
we should have written.

On kms_frontbuffer_tracknig/fbc-2p-primscrn-pri-shrfb-draw-pwrite
(with a 3200x1800 primary screen and a 1920x1080 secondary screen), I
could reduce the runtime from ~7.53s to ~6.01s.

Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
This commit is contained in:
Paulo Zanoni 2015-08-13 17:37:06 -03:00
parent 9194f4efdf
commit 9113c9aa9b
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
lib/igt_draw.c
View File

@ -349,6 +349,7 @@ static void draw_rect_pwrite_tiled(int fd, struct buf_data *buf,
int tmp_used = 0, tmp_size; int tmp_used = 0, tmp_size;
bool flush_tmp = false; bool flush_tmp = false;
int tmp_start_pos = 0; int tmp_start_pos = 0;
int pixels_written = 0;
/* We didn't implement suport for the older tiling methods yet. */ /* We didn't implement suport for the older tiling methods yet. */
igt_require(intel_gen(intel_get_drm_devid(fd)) >= 5); igt_require(intel_gen(intel_get_drm_devid(fd)) >= 5);
@ -380,7 +381,11 @@ static void draw_rect_pwrite_tiled(int fd, struct buf_data *buf,
gem_write(fd, buf->handle, tmp_start_pos, tmp, gem_write(fd, buf->handle, tmp_start_pos, tmp,
tmp_used * pixel_size); tmp_used * pixel_size);
flush_tmp = false; flush_tmp = false;
pixels_written += tmp_used;
tmp_used = 0; tmp_used = 0;
if (pixels_written == rect->w * rect->h)
break;
} }
} }
} }