Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux

drm: Add TODO item for optimizing format helpers

Add a TODO item for optimizing blitting and format-conversion helpers
in DRM and fbdev. There's always demand for faster graphics output.

v4:
* fix typos (Sam)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220223193804.18636-6-tzimmermann@suse.de

+22
+22
Documentation/gpu/todo.rst
··· 241 241 242 242 Level: Advanced 243 243 244 + Benchmark and optimize blitting and format-conversion function 245 + -------------------------------------------------------------- 246 + 247 + Drawing to dispay memory quickly is crucial for many applications' 248 + performance. 249 + 250 + On at least x86-64, sys_imageblit() is significantly slower than 251 + cfb_imageblit(), even though both use the same blitting algorithm and 252 + the latter is written for I/O memory. It turns out that cfb_imageblit() 253 + uses movl instructions, while sys_imageblit apparently does not. This 254 + seems to be a problem with gcc's optimizer. DRM's format-conversion 255 + helpers might be subject to similar issues. 256 + 257 + Benchmark and optimize fbdev's sys_() helpers and DRM's format-conversion 258 + helpers. In cases that can be further optimized, maybe implement a different 259 + algorithm. For micro-optimizations, use movl/movq instructions explicitly. 260 + That might possibly require architecture-specific helpers (e.g., storel() 261 + storeq()). 262 + 263 + Contact: Thomas Zimmermann <tzimmermann@suse.de> 264 + 265 + Level: Intermediate 244 266 245 267 drm_framebuffer_funcs and drm_mode_config_funcs.fb_create cleanup 246 268 -----------------------------------------------------------------