summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* st/dri: Fix setTexBuffer2 with __DRI_TEXTURE_FORMAT_RGB.Chia-I Wu2010-04-081-1/+17
| | | | | When the format is __DRI_TEXTURE_FORMAT_RGB, the texture should be treated as if there is no alpha channel.
* draw llvm: highly reduce the compilation times for draw llvmZack Rusin2010-04-073-67/+56
| | | | | | | | | | our code resets pipe_vertex_buffer's with different offsets when rendering vbo, meaning that we kept creating insane number of shaders even for simple apps e.g. geartrain had 54 shaders and it was taking almost 27 seconds just to compile them. this patch passes pipe_vertex_buffer's to the jit function and lets it to the stride/buffer_offset computation at run time. the slowdown at runtime is largely unnoticable but the we go from 54 shaders to 3, and from 27 seconds to less than 1.
* gallivm: Get the format translation logic write.José Fonseca2010-04-071-5/+19
|
* util: Support fixed formats conversion.José Fonseca2010-04-073-12/+80
|
* util: Add copyright header to u_half.hJosé Fonseca2010-04-071-0/+27
|
* util: Remove the half typedef from p_compiler.h.José Fonseca2010-04-073-8/+6
| | | | Unnecessary, and doesn't even guarantee size.
* llvmpipe: Add missing include.José Fonseca2010-04-071-0/+1
|
* util: Use stubs for the dynamically loaded S3TC functions.José Fonseca2010-04-071-334/+349
| | | | | Loosely based on Luca Barbieri's commit 52e9b990a192a9329006d5f7dd2ac222effea5a5.
* draw: Fix MSVC build (snprintf->util_snprintf).José Fonseca2010-04-071-1/+2
|
* util: Remove u_tile.c YCbCr's.José Fonseca2010-04-072-96/+0
| | | | | Superseded by u_format_yuv.c. Also PIPE_FORMAT_YUYV's interpretation was inconsistent: it was being interpreted as VYUY.
* util: Implement YUV and subsampled RGB format conversion.José Fonseca2010-04-077-7/+1322
|
* draw: init draw->pt.middle.general = NULL just to be safeBrian Paul2010-04-071-0/+2
| | | | | This field should be null from the initial calloc() of the draw context, but let's be safe (and improve understanding of the code).
* llvmpipe: Fix USE_DRAW_LLVM build. Use lp_build_engine.José Fonseca2010-04-073-8/+7
|
* r300: Initialize compiler.max_temp_regs for blits.Michel Dänzer2010-04-071-0/+1
| | | | | | Blits were broken since commit e41a64591bf1a74465bf0adc7d35c991c4cfb4fe ('r300/compiler: make the max number of fragment shader temporaries adjustable').
* draw llvm: Remove unnecessary headers.Vinson Lee2010-04-072-5/+0
|
* gallium: Add files to Makefile.Vinson Lee2010-04-071-1/+4
| | | | | | This was missed in commit ae69f9fbf0a1aab7186e5b644085a5fe5aea99af. New files from the gallium_draw_llvm merge were added to SCons but not make.
* llvmpipe: Remove unused variable.Vinson Lee2010-04-061-1/+0
|
* nouveau: allow multiple simultaneous maps of a pipe_transferBen Skeggs2010-04-071-1/+9
| | | | | | I'm not entirely convinced we want this behaviour (the underlying nouveau_bo doesn't support it either), but since certain parts of the mesa state tracker appear to require it lets make it work for now.
* nv50: fix assertion that caused piglit texturing/texrect-many to failBen Skeggs2010-04-071-1/+1
|
* scons: define DRAW_LLVM if we're building itZack Rusin2010-04-061-0/+2
|
* draw llvm: disable printing of the LLVM irZack Rusin2010-04-061-2/+2
|
* cell: fix buildMarc Dietrich2010-04-062-1/+4
| | | | | | compile fix for cell driver. Signed-off-by: Brian Paul <brianp@vmware.com>
* gallium/draw: add additional DRAW_LLVM checkBrian Paul2010-04-061-0/+3
| | | | Fixes build breakage.
* Merge branch 'gallium_draw_llvm'Zack Rusin2010-04-0616-24/+1934
|\
| * llvmpipe: use a define to decide whether to use draw llvm pathsZack Rusin2010-04-061-1/+5
| | | | | | | | right now disabled by default
| * draw llvm: disable debugging outputZack Rusin2010-04-062-0/+4
| |
| * draw llvm: implement simple pipeline caching using variantsZack Rusin2010-04-063-35/+103
| |
| * draw llvm: fix iteration for larger vertex arraysZack Rusin2010-04-061-8/+8
| | | | | | | | | | | | we were trying to store the outputs starting at the same offset we were using for the input arrays, which was writing beyond the end of the output array.
| * draw llvm: iterate with the correct stop over the outputsZack Rusin2010-04-061-2/+1
| | | | | | | | it's whatever the var step is (4 usually) not an unconditional 1
| * draw llvm: fix draw arraysZack Rusin2010-04-063-11/+14
| | | | | | | | we don't index within the outputs but only within the inputs
| * draw llvm: fix loop iteration and vertex header offsetsZack Rusin2010-04-064-16/+66
| | | | | | | | | | | | the loop was doing a NE comparison which we could have skipped if the prim was triangles (3 verts) and our step was 4 verts. also fix offsets in conversion to aos.
| * draw llvm: when generating the vertex_header struct adjust its nameZack Rusin2010-04-051-1/+4
| | | | | | | | | | change the name to not clash and accuretly represent the number of inputs we store in the data member
| * draw llvm: fix storing of outputs for the rest of the pipelineZack Rusin2010-04-024-54/+94
| | | | | | | | | | | | | | there's no good way of aligning the output's, and since the vertex_header is variable sized in the first place we need to extract elements from a vector and store them individually into an array. this gets the basic examples working again
| * draw llvm: fix translation of formats with variable componentsZack Rusin2010-04-021-0/+13
| |
| * draw llvm: a lot better storing implementationZack Rusin2010-04-013-4/+157
| |
| * Merge remote branch 'origin/master' into gallium_draw_llvmZack Rusin2010-03-31112-2673/+2805
| |\
| * | draw llvm: fix iteration over buffersZack Rusin2010-03-312-4/+26
| | | | | | | | | | | | fetching was converting garbage
| * | Merge remote branch 'origin/master' into gallium_draw_llvmZack Rusin2010-03-30631-7868/+17750
| |\ \
| * | | draw llvm: fix a warningZack Rusin2010-03-301-2/+3
| | | |
| * | | draw llvm: actually set the constant buffers on the contextZack Rusin2010-03-302-2/+8
| | | |
| * | | draw llvm: fix some silly mistakesZack Rusin2010-03-302-8/+15
| | | | | | | | | | | | | | | | | | | | the vs_type selection isn't ideal, but for now both llvmpipe's fs and vs do the same thing which is operate on 4xfloat vector as the base type
| * | | draw llvm: we translate between pointers, not values directlyZack Rusin2010-03-301-11/+12
| | | |
| * | | draw llvm: various fixes for the translation codeZack Rusin2010-03-305-243/+206
| | | | | | | | | | | | | | | | the from translation isn't quite right yet
| * | | draw llvmpipe: lots of fixes for fetch/emitZack Rusin2010-03-294-170/+160
| | | | | | | | | | | | | | | | | | | | the values passed are still not right, but the general scheme is looking good.
| * | | Merge remote branch 'origin/master' into gallium_draw_llvmZack Rusin2010-03-151486-193649/+39091
| |\ \ \
| * | | | gallium/draw: initial code to properly support llvm in the draw moduleZack Rusin2010-02-2212-18/+1601
| | | | | | | | | | | | | | | | | | | | | | | | | code generate big chunks of the vertex pipeline in order to speed up software vertex processing.
* | | | | llvmpipe: Support S3TC when util_format_s3tc_enabled is set.José Fonseca2010-04-061-18/+10
| | | | |
* | | | | gallivm: Fallback to calling util_format_description::fetch_float for any ↵José Fonseca2010-04-064-27/+192
| | | | | | | | | | | | | | | | | | | | format we can't code LLVM IR directly.
* | | | | gallivm: Move the global LLVM objects (module, engine, provider, target) ↵José Fonseca2010-04-065-43/+43
| | | | | | | | | | | | | | | | | | | | into here.
* | | | | util: Respect destination stride in pipe_get_tile_swizzle().Michal Krol2010-04-061-10/+16
| | | | |
generated by cgit v1.1 at 2025-02-10 20:26:02 +0000