summaryrefslogtreecommitdiff
path: root/src/mesa/drivers/dri
AgeCommit message (Collapse)Author
2005-05-31Bugzilla #2195: Convert the radeon driver to the t_vertex interface. This cutsEric Anholt
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>
2005-05-31Move fallback debugging under MACH64_DEBUG=fall (matching other drivers) andEric Anholt
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.
2005-05-30Pull locking out from HW_LOCK/UNLOCK into SpanRenderStart/Finish. Provides aEric Anholt
factor of 5+ improvement in no_rast glxgears and 2.3 in no_rast quake3 (n=1, 320x240, 16bpp, low qualities, 1ghz p3)
2005-05-30Fix software fallbacks by being safe with macro arguments in READ/WRITE_DEPTH.Eric Anholt
Convert to spantmp2.h while I'm here. Tested with 16 and 32bpp no_rast glxgears.
2005-05-30Add the no_rast debug option to mach64. It shows that software fallbacks areEric Anholt
pretty broken.
2005-05-30Pull hardware locking out into SpanRenderStart/Finish rather thanEric Anholt
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.
2005-05-30Add the description of the no_rast option fallback so that MGA_DEBUG=fallEric Anholt
doesn't segfault with no_rast.
2005-05-30Add the no_rast debug option, which I used in some testing.Eric Anholt
2005-05-30Remove unused CTX_ARG2.Eric Anholt
2005-05-28Put some fallback debugging output under DEBUG_FALLBACKS.Eric Anholt
2005-05-28Add a missing EnsureCmdBufSpace, matching r200.Eric Anholt
2005-05-28Add a freebsd-dri-amd64 config, and quiet makedepend slightly by making anEric Anholt
EXPAT_INCLUDES config variable (set on FreeBSD) which is then used in Makefile.template's INCLUDES.
2005-05-28Some fixesJerome Glisse
2005-05-28Up MAX_AOS_ARRAYS to 16. Do we need more?Ben Skeggs
2005-05-28Play with swizzling a bit.Ben Skeggs
2005-05-28Programmable per-vertex point sizes and proper result reg assigns.Aapo Tahkola
2005-05-27More optimal r300Clear.Aapo Tahkola
2005-05-26Bump driver date.Keith Whitwell
2005-05-26constant trackingBen Skeggs
2005-05-26use shared-core rather than sharedDave Airlie
2005-05-26use shared-core instead of sharedDave Airlie
2005-05-25add special case for flat shaded + separate specular trianglesKeith Whitwell
2005-05-25Disable polygon stipple - it seems to break if certain specific stippleKeith Whitwell
patterns are used, including the one generated by conform.
2005-05-25Fix various issues with cliprects and MakeCurrentKeith Whitwell
2005-05-25- Remove one of the loops in emit_arithBen Skeggs
- Handle REPL_ALPHA in emit_arith (possibly incorrect for some things) - Start on getting demos/arbfplight.c to look right. Won't be animated yet, need to re-work const emit so we can update consts without re-translating the entire program. Assertion in r300_state.c::setup_rs_unit needs to be disabled for it to work.
2005-05-24Only use x86 / SSE assembly when USE_X86_ASM / USE_SSE_ASM are defined.Ian Romanick
This makes the driver build on x86-64 again.
2005-05-23Initialize hw stencil renderbuffer.Keith Whitwell
2005-05-23Disable recent sse change pending more testing.Keith Whitwell
2005-05-23Implement enough of ARB_fragment_program for Keith's texenv programBen Skeggs
generation in Mesa. Requires most recent Mesa cvs to work correctly. Swizzle code is possibly slow/buggy and will probably be replaced. Old behaviour can be controlled by a #define in r300_context.h, if no-one minds I'll remove the old code later on. Also, disabled cube map extension. AFAIK We don't support it yet, and it causes issues with UT2004.
2005-05-23Use SSE for vertex emit -- another good speedup.Keith Whitwell
2005-05-23Import Thomas Helstrom's SSE memcpy code from the via X.org driver.Keith Whitwell
Add a TextureMemCpy callback, called from texstore.c when copying texture data via the memcpy_texture() path. Enable this code in the via driver - 100% speedup in texdown.c results.
2005-05-22Small change to handle native case (7 natives case)Jerome Glisse
2005-05-22Adding r300 fragment program swizzle tab. Please someJerome Glisse
one reread macro at begining see if i made any error. A full reread would also be usefull there may be some errors for some swizzle despite the great care i took to done them :)
2005-05-20Add support for modes with a depth-buffer but no stencil-buffer. ThisIan Romanick
eliminates various "3D driver claims to not support visual" messages in 16-bit color mode.
2005-05-20Remove some debug outputJon Smirl
Add #if DEBUG_FRAG in s_nvfragprog.c
2005-05-19Fix EGL fbdev palette problem. demo3 can save ppm of screenJon Smirl
2005-05-19remove the XFree86 ID lineBrian Paul
2005-05-19Ditch unused code and features that arent in subject to get supported any ↵Aapo Tahkola
near time.
2005-05-18In addition to dead-code removal and other clean-up, bump the required DDXIan Romanick
version from 1.0.0 to 1.1.1. 1.1.1 is over a year and a half old at this point, so it should be pretty widespread. This allows the linecomp_sane field to be eliminated from mga_screen_private_s.
2005-05-18Fix inverted y for egl fbdev, add egl demo3Jon Smirl
2005-05-18Stick texture LOD back on.Aapo Tahkola
2005-05-18Polygon mode fixes.Aapo Tahkola
2005-05-18Fix for writing the wrong sysfs variable when blankingJon Smirl
2005-05-18Implement fbdev swapbuffers()Jon Smirl
2005-05-18More comments. More dead-code removed. More clean-up.Ian Romanick
2005-05-17Disable leftover debug statementsKeith Whitwell
2005-05-17Check error return from _eglInitPbufferSurface() correctlyJon Smirl
2005-05-17Compute stride in the right orderJon Smirl
2005-05-17Added software-based pbuffer support (untested).Brian Paul
Use _eglConfigToContextModesRec() function and remove need for the _EGLConfig->glmode struct. Silence some compiler warnings.
2005-05-17Make EGL_NO_MODE_MESA work.Jon Smirl