summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2008-10-07gallium: Introduce PIPE_ARCH_SSE define for SSE support.José Fonseca
Besides meaning x86 and x86-64 architecture, it also depends on SSE2 support enabled on gcc. This fixes the linux-debug build.
2008-10-06Mesa: fix the case where there are no vertex attributes.Stephane Marchesin
2008-10-03CELL: changes to generate SPU code for stencilingRobert Ellison
This set of code changes are for stencil code generation support. Both one-sided and two-sided stenciling are supported. In addition to the raw code generation changes, these changes had to be made elsewhere in the system: - Added new "register set" feature to the SPE assembly generation. A "register set" is a way to allocate multiple registers and free them all at the same time, delegating register allocation management to the spe_function unit. It's quite useful in complex register allocation schemes (like stenciling). - Added and improved SPE macro calculations. These are operations between registers and unsigned integer immediates. In many cases, the calculation can be performed with a single instruction; the macros will generate the single instruction if possible, or generate a register load and register-to-register operation if not. These macro functions are: spe_load_uint() (which has new ways to load a value in a single instruction), spe_and_uint(), spe_xor_uint(), spe_compare_equal_uint(), and spe_compare_greater_uint(). - Added facing to fragment generation. While rendering, the rasterizer needs to be able to determine front- and back-facing fragments, in order to correctly apply two-sided stencil. That requires these changes: - Added front_winding field to the cell_command_render block, so that the state tracker could communicate to the rasterizer what it considered to be the front-facing direction. - Added fragment facing as an input to the fragment function. - Calculated facing is passed during emit_quad().
2008-10-02draw: modify prefetching slightlyKeith Whitwell
2008-10-02draw: don't keep refetching constant inputsKeith Whitwell
2008-10-02rtasm: add prefetch instructionsKeith Whitwell
2008-10-02draw: add streamlined paths for fetching linear vertsKeith Whitwell
2008-10-02mesa: Fix compiler warnings on Windows.Michal Krol
2008-10-02Gallivm: add slt. glxgears should be running, except it isn't.Stephane Marchesin
2008-10-02Gallivm: port to llvm 2.4.Stephane Marchesin
2008-10-02Gallivm: fix off-by-one.Stephane Marchesin
2008-10-02Gallivm: need to link with libstdc++ for llvm.Stephane Marchesin
2008-10-02Gallivm: more instructions.Stephane Marchesin
2008-10-02Gallivm: make it compile again, add some opcodes.Stephane Marchesin
2008-10-02mesa: fix temp register allocation problems.Brian Paul
Complex texcombine modes were running out of registers (>32 registers for 8 tex units).
2008-10-02egl: check for null ptr/nameBrian Paul
2008-10-02egl: remove space after -L flagBrian Paul
2008-10-01Add -msse and -msse2 to the *-x86 configs.José Fonseca
2008-10-01util: No-op u_sse.h outside PIPE_ARCH_X86/X86_64.José Fonseca
2008-10-01tgsi: Include p_config.h.José Fonseca
2008-09-30cell: Fixed usage of MAX_INSTRUCTIONS to use new MAX_PROGRAM_INSTRUCTIONS ↵Jonathan White
instead of old MAX_NV_XXX definitions in order to allow Cell TGSI fragment program generator to work again.
2008-09-30cell: Moved X86 checks to wrap #include section so that Cell targets will ↵Jonathan White
compile again.
2008-10-01util: Fix util_fast_pow/exp2/log2.José Fonseca
- Use a lookup table for log2. - Compute (float) (1 << ipart) by tweaking with the exponent directly to avoid integer overflow and float conversion. - Also table negative exponents to avoid float division and branching. - Implement util_fast_exp as function of util_fast_exp2.
2008-09-30tgsi: SSE2 optimized exp2, log2 and pow implementations.José Fonseca
Special care must be taken when calling compiler generated SSE2 functions from the runtime generated SSE2: saving the xmm registers, and notify gcc the stack is not 16byte aligned. It would be more efficient to keep the stack pointer 16byte aligned, but too hairy, and not consistent in all x86 architectures. This has been tested in linux x86 and windows x86 userspace. Not tested on x86-64 because it is broken for other reasons (even without this change).
2008-09-30util: Header for SSE2 intrinsics portability.José Fonseca
2008-09-29rtasm: Implement immediate group 1 instructions. Fix SIB emition.José Fonseca
2008-09-26cell: checkpoint: more work in emit_function_call()Brian Paul
Simple function call works now, but we don't save/restore the caller's registers yet.
2008-09-26cell: stub-out sin/cos function bodies to avoid trashing caller's stack for nowBrian Paul
2008-09-26gallium: SPU register commentsBrian Paul
2008-09-26cell: move command processing code into new spu_command.c fileBrian Paul
2008-09-26cell: move debug-related declarationsBrian Paul
2008-09-26cell: move debug macros into new spu_debug.hBrian Paul
2008-09-26cell: move really_clear_tiles()Brian Paul
2008-09-26cell: align instruction buffers to 8-byte, not 32-byte boundaryBrian Paul
2008-09-26cell: asst clean-up, var renamingBrian Paul
2008-09-26cell: remove unneeded blend/depth_stencil subclassesBrian Paul
2008-09-26cell: checkpoint: support for function calls in SPU shadersBrian Paul
Will be used for instructions like SIN/COS/POW/TEX/etc. The PPU needs to know the address of some functions in the SPU address space. Send that info to the PPU/main memory rather than patch up shaders on the SPU side. Not finished/tested yet...
2008-09-26cell: inst reorder to save a cycleBrian Paul
2008-09-26st: add prototype for st_get_framebuffer_dimensions()Alan Hourihane
2008-09-26st: export st_get_framebuffer_dimensions() to retrieveAlan Hourihane
the width & height of the fb
2008-09-26mesa: fix incorrect parameter orderBrian Paul
2008-09-26mesa: remove const qualifiers to match device driver interface functionBrian Paul
2008-09-26gallium: rename tgsi_translate_mesa_program() to st_translate_mesa_program()Brian Paul
2008-09-26mesa: add some bracesBrian Paul
2008-09-26mesa: fix/simplify initialization of vertex/fragment program limitsBrian Paul
Defaults for program length, num ALU instructions, num indirections, etc. basically indicate no limit for software rendering. Driver should override as needed.
2008-09-26mesa: increase MAX_INSTRUCTIONSBrian Paul
2008-09-26mesa: fix swizzle failure, fix typoBrian Paul
2008-09-26mesa: fix default buffer object access valueBrian Paul
2008-09-26st: change from ** to * for st_unreference_framebuffer()Alan Hourihane
2008-09-26mesa: fix indentingKeith Whitwell