mirror of
https://gitlab.freedesktop.org/mesa/mesa.git
synced 2026-01-02 13:50:09 +01:00
read-only mirror of https://gitlab.freedesktop.org/mesa/mesa
Changes in v2: - Base class renamed to ir_control_flow_visitor - Tried to comply with coding style This is a new pass that supersedes ir_if_return and "lowers" jumps to if/else structures. Currently it causes no regressions on softpipe and nv40, but I'm not sure whether the piglit glsl tests are thorough enough, so consider this experimental. It can be asked to: 1. Pull jumps out of ifs where possible 2. Remove all "continue"s, replacing them with an "execute flag" 3. Replace all "break" with a single conditional one at the end of the loop 4. Replace all "return"s with a single return at the end of the function, for the main function and/or other functions This gives several great benefits: 1. All functions can be inlined after this pass 2. nv40 and other pre-DX10 chips without "continue" can be supported 3. nv30 and other pre-DX10 chips with no control flow at all are better supported Note that for full effect we should also teach the unroller to unroll loops with a fixed maximum number of iterations but with the canonical conditional "break" that this pass will insert if asked to. Continues are lowered by adding a per-loop "execute flag", initialized to TRUE, that when cleared inhibits all execution until the end of the loop. Breaks are lowered to continues, plus setting a "break flag" that is checked at the end of the loop, and trigger the unique "break". Returns are lowered to breaks/continues, plus adding a "return flag" that causes loops to break again out of their enclosing loops until all the loops are exited: then the "execute flag" logic will ignore everything until the end of the function. Note that "continue" and "return" can also be implemented by adding a dummy loop and using break. However, this is bad for hardware with limited nesting depth, and prevents further optimization, and thus is not currently performed. |
||
|---|---|---|
| bin | ||
| configs | ||
| docs | ||
| doxygen | ||
| include | ||
| scons | ||
| src | ||
| windows | ||
| .emacs-dirvars | ||
| .gitattributes | ||
| .gitignore | ||
| acinclude.m4 | ||
| autogen.sh | ||
| common.py | ||
| configure.ac | ||
| Makefile | ||
| SConstruct | ||
WindML Driver for Mesa 4.0
Requirements
------------
Tornado 2 + WindML, Cumulative Patchs are recommended.
I suppose you have a valid WindML installation. Double buffer hardware
gives better performance than double buffer software so if you can
compile your WindML driver with this option, just do it. I/O
redirection is adviced in target server.
Tested on
---------
During the development, my main target was a CoolMonster:
- Video card: CT69000
- CPU: PENTIUM 266MHz
and my host a Windows NT + Tornado 2.
Installation
------------
1. Mesa sources must be in root directory (C:\)
2. Add the following line to your torVars.bat:
set MESA_BASE=C:\Mesa
OR copy the new torVars.bat in your bin path:
c:/Mesa/src/ugl/tornado/torVars.sample ->
/mnt/nt/Tornado/host/x86-win32/bin/torVars (for example)
3. In a command prompt:
$ torVars
$ cd c:\Mesa
$ make -f Makefile.ugl CPU=PENTIUM
Take a long while...
5. Include all the files from ugldemos folder to build some downloadable
application modules
4. Download UGL/Mesa object files on target
For example via the WindShell:
ld < c:\Tornado\target\lib\objMesaGL.o
ld < c:\Tornado\target\lib\objMesaUGL.o
ld < c:\Tornado\target\lib\objMesaGLU.o
ld < c:\Tornado\target\lib\objGLUTshapes.o
ld < c:\Tornado\target\lib\objMesaOS.o
You can put the previous lines in a file and use:
< filename
6. Download the application modules.
7. In WindShell, run:
-> uglalldemos
During the show some messages will appear, it provides some useful
information on key management.
Coding
------
Sample Usage:
In addition to the usual ugl calls to initialize UGL, (may be find an
input driver), you must do the following to use the UGL/Mesa interface:
1. Call uglMesaCreateContext() to create a UGL/Mesa rendering context,
given the display format.
2. Call uglMesaMakeCurrent() to bind the UGL/Mesa buffers to an
UGL/Mesa Context and to make the context the current one.
3. Make gl* calls to render your graphics.
4. Use uglMesaSwapBuffers() when double buffering to swap front/back buffers.
5. Before the UGL is destroyed, call MesaDestroyContext().
6. Before exiting, call if required uglEventQDestroy and then
uglDeinitialize();
Limitations
-----------
I found the following limitations in my driver :
- Color Indexed management is only in 8 bits
- It's possible to mix UGL/OpenGL application with a software
double buffer
Modifications
------------
New files in Mesa:
- Makefile.ugl
- rules.windmlmesa
- docs/README.UGL
- include/GL/uglmesa.h
- si-glu/Makefile.ugl
- src/Makefile.ugl
- src/ugl/torGLUTShapesInit.c
- src/ugl/torMesaUGLInit.c
- src/ugl/ugl_api.c
- src/ugl/ugl_dd.c
- src/ugl/ugl_glutshapes.c
- src/ugl/ugl_line.c
- src/ugl/ugl_span.c
- src/ugl/ugl_tri.c
- src/ugl/uglmesaP.h
- ugldemos/*
Modified files in Tornado 2.0:
- c:\Tornado\host\x86-win32\bin\torVars.bat
rem Command line build environments
set WIND_HOST_TYPE=x86-win32
set WIND_BASE=C:\Tornado
set MESA_BASE=C:\Mesa
set PATH=%WIND_BASE%\host\%WIND_HOST_TYPE%\bin;%PATH%
- c:\Tornado\target\config\comps\VxWorks\01uglmesa.cdf
- c:\Tornado\target\h\GL\*
Todo
----
- GCC 2.96, ASM compilation
Thanks to:
----------
Precision Insight team for their great job around Mesa, XFree, and DRI.
Wind River Systems to take me as an intern.
Stephane Raimbault
<stephane.raimbault@windriver.com>
<stephane.raimbault@deesse.univ-lemans.fr>
July 24, 2001