Keith Whitwell [Fri, 10 Jun 2005 08:32:27 +0000 (08:32 +0000)]
Add notes about intended precision of opcodes.
Remove dead floating point twiddles.
Fix bug translating MAD->MUL,ADD.
Keith Whitwell [Thu, 9 Jun 2005 22:10:23 +0000 (22:10 +0000)]
Fix a couple of buglets
Keith Whitwell [Thu, 9 Jun 2005 19:03:18 +0000 (19:03 +0000)]
Another typo
Keith Whitwell [Thu, 9 Jun 2005 18:59:41 +0000 (18:59 +0000)]
Further clean up RoughApproxPow2 and disable the optimized version --
it's a very poor approximation.
Keith Whitwell [Thu, 9 Jun 2005 18:52:42 +0000 (18:52 +0000)]
Fix typo in RoughApproxPow2
Keith Whitwell [Thu, 9 Jun 2005 14:55:34 +0000 (14:55 +0000)]
Store compiled vertex program representations in a pointer in the
vertex_program struct.
Allow switching between regular and vertex_program implementations
of fixed function TNL with the MESA_TNL_PROG environment var
(previously this required recompilation).
Ensure program compilation only references program data, not the
wider context. This means that compiled programs only need to be
invalidated when the program string changes, not on other state
changes.
Keith Whitwell [Thu, 9 Jun 2005 14:52:08 +0000 (14:52 +0000)]
Build vertex programs using state from a 'state key' struct which is
distilled from the current state.
Maintain a list of vertex programs for various different tnl states
indexed by these state keys.
This cuts down the number of vertex programs generated and means that
time spent optimizing vertex programs for execution isn't wasted each
time the state changes.
Keith Whitwell [Wed, 8 Jun 2005 22:14:29 +0000 (22:14 +0000)]
Turn off debug
Keith Whitwell [Wed, 8 Jun 2005 22:10:34 +0000 (22:10 +0000)]
Add t_vb_arbprogram_sse.c
Keith Whitwell [Wed, 8 Jun 2005 22:10:16 +0000 (22:10 +0000)]
Runtime generate sse/sse2 code for some vertex programs. Experimental
and only active when _TNL_FIXED_FUNCTION_PROGRAM is defined *and*
the MESA_EXPERIMENTAL environment variable is set...
Works for the arbfplight demo at least.
Keith Whitwell [Wed, 8 Jun 2005 22:06:29 +0000 (22:06 +0000)]
Turn off DISASSEM
Keith Whitwell [Wed, 8 Jun 2005 21:58:30 +0000 (21:58 +0000)]
New arguments for sse_shufps()
Keith Whitwell [Wed, 8 Jun 2005 21:57:45 +0000 (21:57 +0000)]
Use ALIGN_MALLOC for parameter lists.
Keith Whitwell [Wed, 8 Jun 2005 21:56:21 +0000 (21:56 +0000)]
Add some more opcodes
Ben Skeggs [Wed, 8 Jun 2005 15:05:19 +0000 (15:05 +0000)]
- Secondary colour
- Calculate tc_count in EmitArrays (enabled units != nr texcoords).
Brian Paul [Wed, 8 Jun 2005 13:35:28 +0000 (13:35 +0000)]
regenerated
Brian Paul [Wed, 8 Jun 2005 13:35:07 +0000 (13:35 +0000)]
put full prototype on one line to fix gl_mangle.h generation (Dave Reveman)
Brian Paul [Wed, 8 Jun 2005 13:07:33 +0000 (13:07 +0000)]
remove stray line
Keith Whitwell [Wed, 8 Jun 2005 10:03:29 +0000 (10:03 +0000)]
Remove ARL opcode from interpreter.
Revert SEL opcode to MSK.
Split out t_vb_arbprogram.h header file.
Keith Whitwell [Wed, 8 Jun 2005 10:02:28 +0000 (10:02 +0000)]
new file
Brian Paul [Wed, 8 Jun 2005 02:18:22 +0000 (02:18 +0000)]
regenerate for latest glext.h
Brian Paul [Wed, 8 Jun 2005 02:17:10 +0000 (02:17 +0000)]
fix some renderbuffer problems for Xserver build (Dave Reveman)
Brian Paul [Wed, 8 Jun 2005 02:14:55 +0000 (02:14 +0000)]
check for double-buffered buffer in XMesaDestroyBuffer() (Dave Reveman)
Daniel Borca [Tue, 7 Jun 2005 19:01:33 +0000 (19:01 +0000)]
updates from Paul Garceau
Brian Paul [Tue, 7 Jun 2005 15:30:12 +0000 (15:30 +0000)]
added GL libs to APP_LIB_DEPS (bug 3485)
Brian Paul [Tue, 7 Jun 2005 15:25:08 +0000 (15:25 +0000)]
fix error in OSMESA_LIB_NAME line
Brian Paul [Tue, 7 Jun 2005 15:03:40 +0000 (15:03 +0000)]
fix GL_ALPHA render-to-texture problem (Jon Smirl)
Keith Whitwell [Tue, 7 Jun 2005 12:46:30 +0000 (12:46 +0000)]
Initialize some variables.
Keith Whitwell [Tue, 7 Jun 2005 12:45:29 +0000 (12:45 +0000)]
Add new runtime assembly files.
Keith Whitwell [Tue, 7 Jun 2005 12:45:04 +0000 (12:45 +0000)]
Split out runtime assembly functions.
Keith Whitwell [Tue, 7 Jun 2005 12:44:26 +0000 (12:44 +0000)]
New files - split off runtime assembly functions from
t_vertex_sse.c.
Keith Whitwell [Tue, 7 Jun 2005 10:59:37 +0000 (10:59 +0000)]
Simplify interpreted language:
- Expand operand argument specifiers so that all registers and
parameters can be referenced directly. Remove old PAR/PRL
instructions.
- No 3 operand instructions, translate MAD -> MUL, ADD. (No room for
3 full operands in instruction struct).
- Translate SWZ instructions into 1 or 2 reduced swizzles at compile
time.
- Add hardwired code for moving input and output values to/from the
register file. Drop old INx, OUT instructions.
Keith Whitwell [Tue, 7 Jun 2005 10:56:17 +0000 (10:56 +0000)]
Add a VP_MAX_OPCODE entry to allow engines with internal opcodes to
place them sensibly.
Daniel Borca [Tue, 7 Jun 2005 05:29:38 +0000 (05:29 +0000)]
updates from Paul Garceau
Jon Smirl [Tue, 7 Jun 2005 02:10:14 +0000 (02:10 +0000)]
Add SetBuffer to egl driver too
Jon Smirl [Mon, 6 Jun 2005 21:40:55 +0000 (21:40 +0000)]
Add a NULL SetBuffer function to the fbdev driver
Brian Paul [Mon, 6 Jun 2005 14:40:56 +0000 (14:40 +0000)]
this is auto-generated
Ben Skeggs [Mon, 6 Jun 2005 07:38:29 +0000 (07:38 +0000)]
TEX to output
Ben Skeggs [Sun, 5 Jun 2005 08:25:54 +0000 (08:25 +0000)]
Fix xyz/w interaction (needs a cleanup still..)
Use SRC0A instead of WZY/XXX combination for W in XYZ positions.
Remove dodgy hack from POW opcode, now works correctly without it
Jon Smirl [Sun, 5 Jun 2005 02:10:58 +0000 (02:10 +0000)]
Make install install GLES too
Brian Paul [Sat, 4 Jun 2005 22:09:27 +0000 (22:09 +0000)]
set OLD_RENDERBUFFER=0, need to clean up a few more things before totally removing that
Brian Paul [Sat, 4 Jun 2005 18:25:39 +0000 (18:25 +0000)]
remove old span functions from swrast_device_driver struct
Brian Paul [Sat, 4 Jun 2005 18:12:28 +0000 (18:12 +0000)]
remove old span code
Brian Paul [Sat, 4 Jun 2005 17:53:30 +0000 (17:53 +0000)]
fix an assertion
Brian Paul [Sat, 4 Jun 2005 17:45:50 +0000 (17:45 +0000)]
remove old span code
Brian Paul [Sat, 4 Jun 2005 17:44:42 +0000 (17:44 +0000)]
added comment
Karl Schultz [Fri, 3 Jun 2005 15:24:55 +0000 (15:24 +0000)]
Fixup file lists for tarball/zip file creation.
- Adjust for recent changes to Windows build files.
- Add in at least some of the new slang files - enough to build on
Windows, but some of the other slang-related files may still need to be
added. (Will probably get corrected just prior to next release, if needed)
Karl Schultz [Thu, 2 Jun 2005 23:17:02 +0000 (23:17 +0000)]
Updates for new Windows build system.
Brian Paul [Thu, 2 Jun 2005 04:09:05 +0000 (04:09 +0000)]
Remove last of the old span code.
Brian Paul [Thu, 2 Jun 2005 04:07:08 +0000 (04:07 +0000)]
oops, undo previous change
Brian Paul [Thu, 2 Jun 2005 04:05:45 +0000 (04:05 +0000)]
start removing old (pre-renderbuffer) span code
Jouk Jansen [Wed, 1 Jun 2005 12:07:15 +0000 (12:07 +0000)]
Committing in .
OpenVMS makefile added
Added Files:
Mesa/src/mesa/shader/grammar/descrip.mms
----------------------------------------------------------------------
Jouk Jansen [Wed, 1 Jun 2005 12:05:34 +0000 (12:05 +0000)]
Committing in .
Update OpneVMS compilesupport
Modified Files:
Mesa/src/mesa/shader/descrip.mms
Mesa/src/mesa/shader/slang/descrip.mms
Mesa/src/mesa/shader/slang/slang_compile.c
Mesa/src/mesa/shader/slang/slang_storage.c
Mesa/src/mesa/shader/slang/slang_storage.h
----------------------------------------------------------------------
Eric Anholt [Wed, 1 Jun 2005 08:22:57 +0000 (08:22 +0000)]
Move fallback and other debugging under TDFX_DEBUG using the same mechanism
as other drivers.
Eric Anholt [Wed, 1 Jun 2005 07:44:52 +0000 (07:44 +0000)]
Add no_rast option.
Keith Whitwell [Wed, 1 Jun 2005 07:02:23 +0000 (07:02 +0000)]
Add a mode to test flat shading + separate specular.
Eric Anholt [Wed, 1 Jun 2005 02:59:12 +0000 (02:59 +0000)]
Fix the 3dfx driver by correcting what appears to be a typo GL_RGBA8 instead
of GL_RGBA in driNewRenderbuffer of the backbuffer.
Brian Paul [Wed, 1 Jun 2005 00:50:23 +0000 (00:50 +0000)]
added code to detect GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT error
Brian Paul [Tue, 31 May 2005 23:55:21 +0000 (23:55 +0000)]
add support for GL_RENDERBUFFER_RED/GREEN/BLUE/etc_SIZE_EXT querie
Brian Paul [Tue, 31 May 2005 23:54:44 +0000 (23:54 +0000)]
remove temporary GL_EXT_framebuffer_object info
Brian Paul [Tue, 31 May 2005 23:54:24 +0000 (23:54 +0000)]
updated to version 28
Eric Anholt [Tue, 31 May 2005 23:32:47 +0000 (23:32 +0000)]
Bugzilla #2708: Disable the fallback on GL_SPHERE_MAP. It appears to hurt more
than it helps, and we seem to have a handle on how to fix it.
Alan Hourihane [Tue, 31 May 2005 18:10:03 +0000 (18:10 +0000)]
Add Intel(R) 945G support (Keith Whitwell, Tungsten Graphics)
Aapo Tahkola [Tue, 31 May 2005 05:53:38 +0000 (05:53 +0000)]
Causes lockups on some rare cases.
Eric Anholt [Tue, 31 May 2005 05:31:43 +0000 (05:31 +0000)]
Pull locking out of HW_LOCK/UNLOCK and into SpanRenderStart/Finish. Also do
some cleanups on the span macros. This does not deal with the brokenness of
depth spans, but makes the code more readable.
Eric Anholt [Tue, 31 May 2005 04:26:13 +0000 (04:26 +0000)]
Correct the descriptions of the glBlend{Equation,Func} fallbacks in r128, and
remove those descriptions and the corresponding bits on r200, where they're no
longer used.
Eric Anholt [Tue, 31 May 2005 04:13:45 +0000 (04:13 +0000)]
Remove unused CTX_ARG2 define.
Eric Anholt [Tue, 31 May 2005 04:04:24 +0000 (04:04 +0000)]
Bugzilla #2195: Convert the radeon driver to the t_vertex interface. This cuts
about 200 lines from the code and 25k from the binary, while matching other
drivers more closely. In the worst case (tcl_mode=0) it appears to have
a performance cost of 4.4% +/- 0.3% on quake3 (800x600 demofours, 1ghz p3,
rv200). Tested on ut2004, ut, q3, projtex.
Submitted by: Andreas Stenglein <a.stenglein@gmx.net>
Eric Anholt [Tue, 31 May 2005 03:47:39 +0000 (03:47 +0000)]
Move fallback debugging under MACH64_DEBUG=fall (matching other drivers) and
add pretty descriptions of the bits. GC one of the bits that was unused.
Also, now only the first thing triggering a fallback and the last thing
requiring one do the debug output. This also matches other drivers.
Eric Anholt [Mon, 30 May 2005 08:07:03 +0000 (08:07 +0000)]
Pull locking out from HW_LOCK/UNLOCK into SpanRenderStart/Finish. Provides a
factor of 5+ improvement in no_rast glxgears and 2.3 in no_rast quake3 (n=1,
320x240, 16bpp, low qualities, 1ghz p3)
Eric Anholt [Mon, 30 May 2005 06:50:17 +0000 (06:50 +0000)]
Fix software fallbacks by being safe with macro arguments in READ/WRITE_DEPTH.
Convert to spantmp2.h while I'm here. Tested with 16 and 32bpp no_rast
glxgears.
Eric Anholt [Mon, 30 May 2005 06:41:19 +0000 (06:41 +0000)]
Add the no_rast debug option to mach64. It shows that software fallbacks are
pretty broken.
Eric Anholt [Mon, 30 May 2005 06:15:29 +0000 (06:15 +0000)]
Pull hardware locking out into SpanRenderStart/Finish rather than
HW_LOCK/UNLOCK. Provides a 61% performance improvement in no_rast=true quake3
(n=2, 320x240, 16bpp, p4 1ghz), and a factor of 3 improvement in no_rast
glxgears.
Eric Anholt [Mon, 30 May 2005 06:11:22 +0000 (06:11 +0000)]
Add the description of the no_rast option fallback so that MGA_DEBUG=fall
doesn't segfault with no_rast.
Eric Anholt [Mon, 30 May 2005 03:32:52 +0000 (03:32 +0000)]
Add the no_rast debug option, which I used in some testing.
Eric Anholt [Mon, 30 May 2005 01:59:39 +0000 (01:59 +0000)]
Remove unused CTX_ARG2.
Eric Anholt [Sat, 28 May 2005 23:45:39 +0000 (23:45 +0000)]
Put some fallback debugging output under DEBUG_FALLBACKS.
Eric Anholt [Sat, 28 May 2005 22:56:16 +0000 (22:56 +0000)]
Add a missing EnsureCmdBufSpace, matching r200.
Eric Anholt [Sat, 28 May 2005 20:17:06 +0000 (20:17 +0000)]
Add a freebsd-dri-amd64 config, and quiet makedepend slightly by making an
EXPAT_INCLUDES config variable (set on FreeBSD) which is then used in
Makefile.template's INCLUDES.
Jerome Glisse [Sat, 28 May 2005 12:18:32 +0000 (12:18 +0000)]
Some fixes
Dave Airlie [Sat, 28 May 2005 10:55:23 +0000 (10:55 +0000)]
use shared-core
Ben Skeggs [Sat, 28 May 2005 05:20:05 +0000 (05:20 +0000)]
Up MAX_AOS_ARRAYS to 16. Do we need more?
Ben Skeggs [Sat, 28 May 2005 03:19:56 +0000 (03:19 +0000)]
Play with swizzling a bit.
Aapo Tahkola [Sat, 28 May 2005 01:02:00 +0000 (01:02 +0000)]
Programmable per-vertex point sizes and proper result reg assigns.
Aapo Tahkola [Fri, 27 May 2005 20:10:05 +0000 (20:10 +0000)]
More optimal r300Clear.
Brian Paul [Fri, 27 May 2005 14:44:14 +0000 (14:44 +0000)]
properly pop the GL_TEXTURE_GEN_ enable state
Karl Schultz [Thu, 26 May 2005 18:22:13 +0000 (18:22 +0000)]
Account for moved/new/removed files.
Ian Romanick [Thu, 26 May 2005 16:59:47 +0000 (16:59 +0000)]
Fix gl_enums.py so that enums that are listed with the same name in
multiple places don't get generated multiple times.
Ian Romanick [Thu, 26 May 2005 16:34:58 +0000 (16:34 +0000)]
Minor correction to EXT_framebuffer_object XML. Fix gl_table.py so that it
generates the correct script name in the generated output.
Ian Romanick [Thu, 26 May 2005 16:24:35 +0000 (16:24 +0000)]
Make a couple minor corrections to gl_API.xml. Fixes the name of an
extension that was missing the GL_ part and uses the core GL type names for
vertex buffer object functions instead of the ARB names.
Also commits the resulting changes to the generated code. Some how, the
#if sequence disappeared in the 1.52 version of glapi_x86.S.
Keith Whitwell [Thu, 26 May 2005 11:42:19 +0000 (11:42 +0000)]
Bump driver date.
Dave Airlie [Thu, 26 May 2005 04:24:09 +0000 (04:24 +0000)]
make egl build first so we don't get wierd drivers linking to libs not available
yet
Dave Airlie [Thu, 26 May 2005 03:47:34 +0000 (03:47 +0000)]
make linux-solo and linux-solo-x86 config files hierarchical
Ben Skeggs [Thu, 26 May 2005 02:33:57 +0000 (02:33 +0000)]
constant tracking
Dave Airlie [Thu, 26 May 2005 01:40:34 +0000 (01:40 +0000)]
use shared-core rather than shared
Dave Airlie [Thu, 26 May 2005 01:40:00 +0000 (01:40 +0000)]
use shared-core instead of shared
Ian Romanick [Wed, 25 May 2005 21:00:16 +0000 (21:00 +0000)]
Fix a couple typeos in enum names.
Michal Krol [Wed, 25 May 2005 15:09:32 +0000 (15:09 +0000)]
change local var names from "asm" to "assem" - I guess it caused
gcc 3.3.5 to break, but not tested it
Michal Krol [Wed, 25 May 2005 14:58:22 +0000 (14:58 +0000)]
a small utility to convert .syn files to its binary forms;
does not require any command line params;
Michal Krol [Wed, 25 May 2005 14:27:59 +0000 (14:27 +0000)]
silencium gcc warnings