fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-03-10 22:20:40 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

docs/perf: Drop a bunch of old hints on getting faster GL1 swrast.

Browse source 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39741>
This commit is contained in:
Emma Anholt 2026-02-05 18:23:00 -08:00
parent f56d0faca9
commit 9fbd722e47
1 changed files with 0 additions and 43 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

43
docs/perf.rst
View file

@ -24,46 +24,3 @@ Intel GPUs
.. _ReBAR: https://www.intel.com/content/www/us/en/support/articles/000090831/graphics.html
.. _ASPM: https://docs.redhat.com/en/documentation/red_hat_enterprise_linux/6/html/power_management_guide/aspm
.. _cpufreq: https://docs.redhat.com/en/documentation/red_hat_enterprise_linux/6/html/performance_tuning_guide/s-cpu-cpufreq
Software rendering
------------------
#. Turn off smooth shading when you don't need it (glShadeModel)
#. Turn off depth buffering when you don't need it.
#. Turn off dithering when not needed.
#. Use double buffering as it's often faster than single buffering
#. Compile in the X Shared Memory extension option if it's supported on
your system by adding -DSHM to CFLAGS and -lXext to XLIBS for your
system in the Make-config file.
#. Recompile Mesa with more optimization if possible.
#. Try to maximize the amount of drawing done between glBegin/glEnd
pairs.
#. Use the MESA_BACK_BUFFER variable to find best performance in double
buffered mode. (X users only)
#. Optimized polygon rasterizers are employed when: rendering into back
buffer which is an XImage RGB mode, not grayscale, not monochrome
depth buffering is GL_LESS, or disabled flat or smooth shading
dithered or non-dithered no other rasterization operations enabled
(blending, stencil, etc)
#. Optimized line drawing is employed when: rendering into back buffer
which is an XImage RGB mode, not grayscale, not monochrome depth
buffering is GL_LESS or disabled flat shading dithered or
non-dithered no other rasterization operations enabled (blending,
stencil, etc)
#. Textured polygons are fastest when: using a 3-component (RGB), 2-D
texture minification and magnification filters are GL_NEAREST texture
coordinate wrap modes for S and T are GL_REPEAT GL_DECAL environment
mode glHint( GL_PERSPECTIVE_CORRECTION_HINT, GL_FASTEST ) depth
buffering is GL_LESS or disabled
#. Lighting is fastest when: Two-sided lighting is disabled
GL_LIGHT_MODEL_LOCAL_VIEWER is false GL_COLOR_MATERIAL is disabled No
spot lights are used (all GL_SPOT_CUTOFFs are 180.0) No local lights
are used (all position W's are 0.0) All material and light
coefficients are >= zero
#. XFree86 users: if you want to use 24-bit color try starting your X
server in 32-bit per pixel mode for better performance. That is,
start your X server with startx -- -bpp 32 instead of startx -- -bpp
24
#. Try disabling dithering with the MESA_NO_DITHER environment variable.
If this environment variable is defined Mesa will disable dithering
and the command glEnable(GL_DITHER) will be ignored.