summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* i965: Remove redundant discard jumps.Kenneth Graunke2015-02-242-0/+43
| | | | | | | | | | | | | | | | | | | | | | | | | | With the previous optimization in place, some shaders wind up with multiple discard jumps in a row, or jumps directly to the next instruction. We can remove those. Without NIR on Haswell: total instructions in shared programs: 5777258 -> 5775872 (-0.02%) instructions in affected programs: 20312 -> 18926 (-6.82%) helped: 716 With NIR on Haswell: total instructions in shared programs: 5773163 -> 5771785 (-0.02%) instructions in affected programs: 21040 -> 19662 (-6.55%) helped: 717 v2: Use the CFG rather than the old instructions list. Presumably the placeholder halt will be in the last basic block. v3: Make sure placeholder_halt->prev isn't the head sentinel (caught twice by Eric Anholt). Signed-off-by: Kenneth Graunke <[email protected]> Reviewed-by: Matt Turner <[email protected]> Reviewed-by: Eric Anholt <[email protected]>
* glsl: Optimize "if (cond) discard;" to a conditional discard.Kenneth Graunke2015-02-244-0/+84
| | | | | | | | | | | | | | | | | | | | | | | | | | st_glsl_to_tgsi and ir_to_mesa have handled conditional discards for a long time; the previous patch added that capability to i965. i965 (Haswell) shader-db stats: Without NIR: total instructions in shared programs: 5792133 -> 5776360 (-0.27%) instructions in affected programs: 737585 -> 721812 (-2.14%) helped: 6300 HURT: 68 GAINED: 2 With NIR: total instructions in shared programs: 5787538 -> 5769569 (-0.31%) instructions in affected programs: 767843 -> 749874 (-2.34%) helped: 6522 HURT: 35 GAINED: 6 Signed-off-by: Kenneth Graunke <[email protected]> Reviewed-by: Connor Abbott <[email protected]> Reviewed-by: Matt Turner <[email protected]> Reviewed-by: Eric Anholt <[email protected]>
* i965/fs: Handle conditional discards.Kenneth Graunke2015-02-242-17/+26
| | | | | | | | | | | | | The discard condition tells us which channels we want killed. We want to invert that condition to get the channels that should survive (remain live) in f0.1. Emit a CMP to negate it. Nothing generates these today, but that will change shortly. Signed-off-by: Kenneth Graunke <[email protected]> Reviewed-by: Connor Abbott <[email protected]> Reviewed-by: Matt Turner <[email protected]> Reviewed-by: Eric Anholt <[email protected]>
* nir: Introduce nir_intrinsic_discard_if.Kenneth Graunke2015-02-242-2/+10
| | | | | | | | | | | | This is a conditional discard, which takes a boolean source. Note that we don't generate ir_discard::condition today, so this shouldn't break drivers (since none implement this intrinsic yet). Signed-off-by: Kenneth Graunke <[email protected]> Reviewed-by: Connor Abbott <[email protected]> Reviewed-by: Matt Turner <[email protected]> Reviewed-by: Eric Anholt <[email protected]>
* glsl: Delete dead discard conditions in constant folding.Kenneth Graunke2015-02-241-0/+24
| | | | | | | | | | | | | | | | opt_constant_folding() already detects conditional assignments where the condition is constant, and either deletes the assignment or the condition. Make it handle discards in the same fashion. Spotted happening in the wild in Tropico 5 shaders. Signed-off-by: Kenneth Graunke <[email protected]> Reviewed-by: Ian Romanick <[email protected]> Reviewed-by: Connor Abbott <[email protected]> Reviewed-by: Matt Turner <[email protected]> Reviewed-by: Eric Anholt <[email protected]>
* glsl: Handle conditional discards in lower_discard_flow().Kenneth Graunke2015-02-241-1/+8
| | | | | | | | | | | | | This pass wasn't prepared to handle conditional discards. Instead of initializing the "discarded" temporary to "true", set it to the condition. Then, refer to the variable for the condition, to avoid duplicating the expression tree. Signed-off-by: Kenneth Graunke <[email protected]> Reviewed-by: Connor Abbott <[email protected]> Reviewed-by: Matt Turner <[email protected]> Reviewed-by: Eric Anholt <[email protected]>
* glsl: Make ir_rvalue_visitor visit ir_discard::condition.Kenneth Graunke2015-02-242-0/+22
| | | | | | | | | | | | | This was forgotten. I omitted the NULL check since we don't check ir_assignment::condition either. Signed-off-by: Kenneth Graunke <[email protected]> Reviewed-by: Ian Romanick <[email protected]> Reviewed-by: Connor Abbott <[email protected]> Reviewed-by: Matt Turner <[email protected]> Reviewed-by: Eric Anholt <[email protected]>
* glsl: Make ir_validate check the type of ir_discard::condition.Kenneth Graunke2015-02-241-0/+15
| | | | | | | | | | Copy and pasted from the ir_if::condition handling, plus a NULL check. Signed-off-by: Kenneth Graunke <[email protected]> Reviewed-by: Ian Romanick <[email protected]> Reviewed-by: Connor Abbott <[email protected]> Reviewed-by: Matt Turner <[email protected]> Reviewed-by: Eric Anholt <[email protected]>
* Revert "i965/fs: Remove force_writemask_all assertion for execsize < 8."Matt Turner2015-02-241-0/+1
| | | | | | | | | This reverts commit 0d8f27eab7b7e8b7a16e76aabd3f6a0ab4880497. "This doesn't seem to be necessary." <- I was wrong! Tested-by: Markus Wick <[email protected]> Reviewed-by: Francisco Jerez <[email protected]>
* i965/fs: Emit MOV(1) instructions with force_writemask_all.Matt Turner2015-02-241-0/+1
| | | | | | | Fixes rendering with Dolphin. Tested-by: Markus Wick <[email protected]> Reviewed-by: Francisco Jerez <[email protected]>
* i965/fs: Optimize (gl_FrontFacing ? x : y) where x and y are ±1.0.Matt Turner2015-02-242-0/+95
| | | | | | | total instructions in shared programs: 5695356 -> 5689775 (-0.10%) instructions in affected programs: 486231 -> 480650 (-1.15%) helped: 2604 LOST: 1
* i965/fs/nir: Optimize integer multiply by a 16-bit constant.Matt Turner2015-02-241-1/+23
| | | | | | | | | | | | Gen8+ support was just broken, since MUL now consumes 32-bits from both sources. Fixes 986 piglit tests on my BDW. total instructions in shared programs: 7753873 -> 7753522 (-0.00%) instructions in affected programs: 28164 -> 27813 (-1.25%) helped: 77 GAINED: 47 Reviewed-by: Ian Romanick <[email protected]>
* i965/fs/nir: Optimize (gl_FrontFacing ? x : y) where x and y are ±1.0.Matt Turner2015-02-242-0/+90
| | | | | | | | total instructions in shared programs: 7756214 -> 7753873 (-0.03%) instructions in affected programs: 455452 -> 453111 (-0.51%) helped: 2333 Reviewed-by: Eric Anholt <[email protected]>
* nir/register: Add a parent_instr fieldJason Ekstrand2015-02-243-1/+18
| | | | | | | | | | | | | This adds a parent_instr field similar to the one for ssa_def. The difference here is that the parent_instr field on a nir_register can be NULL if the register does not have a unique definition or if that definition does not dominate all its uses. We set this field in the out-of-SSA pass so that backends can get SSA-like information even after they have gone out of SSA. Reviewed-by: Connor Abbott <[email protected]> Reviewed-by: Matt Turner <[email protected]> Reviewed-by: Eric Anholt <[email protected]>
* st/mesa: remove unused/broken function st_print_shadersMarek Olšák2015-02-242-49/+0
| | | | Reviewed-by: Dave Airlie <[email protected]>
* st/mesa: remove struct qualifier from st_src_reg parameterBrian Paul2015-02-241-1/+1
| | | | It's a class. Silences MSVC warning.
* mesa: remove INV_SQRTF() macroBrian Paul2015-02-2410-22/+15
| | | | | Reviewed-by: Roland Scheidegger <[email protected]> Reviewed-by: Matt Turner <[email protected]>
* mesa: remove ceilf, floorf macrosBrian Paul2015-02-241-2/+0
| | | | Reviewed-by: Matt Turner <[email protected]>
* mesa: remove expf macroBrian Paul2015-02-241-1/+0
| | | | Reviewed-by: Matt Turner <[email protected]>
* mesa: remove logf macroBrian Paul2015-02-242-1/+1
| | | | Reviewed-by: Matt Turner <[email protected]>
* mesa: remove powf macroBrian Paul2015-02-244-2/+3
| | | | | | Use the wrapper in c99_math.h if needed. Reviewed-by: Matt Turner <[email protected]>
* mesa: remove unused exp2f, log2f, truncf wrappersBrian Paul2015-02-241-10/+0
| | | | Reviewed-by: Matt Turner <[email protected]>
* mesa: remove unused acosf, asinf, atan2f, etc. macrosBrian Paul2015-02-241-16/+0
| | | | | | | Not used anywhere. If any of these are needed, they should be added to c99_math.h Reviewed-by: Matt Turner <[email protected]>
* mesa: replace FABSF with fabsfBrian Paul2015-02-2411-53/+48
| | | | Reviewed-by: Matt Turner <[email protected]>
* mesa: replace FLOORF with floorfBrian Paul2015-02-243-13/+11
| | | | Reviewed-by: Matt Turner <[email protected]>
* mesa: remove unused CEILF macroBrian Paul2015-02-241-3/+0
| | | | Reviewed-by: Matt Turner <[email protected]>
* mesa: replace LOGF, EXPF with logf, expfBrian Paul2015-02-245-12/+18
| | | | Reviewed-by: Matt Turner <[email protected]>
* mesa: replace FREXPF, LDEXPF with frexpf, ldexpfBrian Paul2015-02-244-11/+20
| | | | | | Start getting rid of some imports.h macros. Use the c99 functions instead. Reviewed-by: Matt Turner <[email protected]>
* targets/libgl-xlib: add src/ include dir to fix buildBrian Paul2015-02-241-0/+1
|
* swrast: fix a few release build warningsBrian Paul2015-02-242-0/+2
|
* r600g,radeonsi: fix streamout after pipeline stats have been usedMarek Olšák2015-02-242-13/+1
| | | | | | | | | | | EVENT_TYPE_PIPELINESTAT_STOP disables streamout queries too. Luckily, pipeline stats are enabled by default, so we don't even have to emit EVENT_TYPE_PIPELINESTAT_START. Tested on Hawaii, Bonaire, Redwood, RV730. Reviewed-by: Michel Dänzer <[email protected]>
* radeonsi: small cleanup around current_rast_primMarek Olšák2015-02-241-9/+6
| | | | | | | | - remove the last parameter of si_emit_rasterizer_prim_state - remove the last unused parameter of si_emit_draw_registers - use current_rast_prim in si_emit_draw_registers Reviewed-by: Michel Dänzer <[email protected]>
* radeonsi: set current_rast_prim in the right placeMarek Olšák2015-02-241-1/+7
| | | | Reviewed-by: Michel Dänzer <[email protected]>
* radeonsi: simplify obtaining a shader property in si_emit_clip_regsMarek Olšák2015-02-241-2/+1
| | | | Reviewed-by: Michel Dänzer <[email protected]>
* radeonsi: only preload VertexID for the GS copy shaderMarek Olšák2015-02-241-4/+6
| | | | | | The copy shader doesn't use any other preloaded VGPRs. Reviewed-by: Michel Dänzer <[email protected]>
* radeonsi: dump the shader key when dumping shadersMarek Olšák2015-02-241-0/+38
| | | | Reviewed-by: Michel Dänzer <[email protected]>
* r600g,radeonsi: cleanup of hex literalsMarek Olšák2015-02-244-32/+32
| | | | | | | | 0x3F800000 -> fui(1.0) 0x00000000 -> 0 Reviewed-by: Dave Airlie <[email protected]> Reviewed-by: Michel Dänzer <[email protected]>
* radeonsi: set PA_SU_HARDWARE_SCREEN_OFFSET to 0Marek Olšák2015-02-241-0/+2
| | | | | | It was probably 0 already, but it doesn't hurt to set it. Reviewed-by: Michel Dänzer <[email protected]>
* st/mesa: cleanup st_translate_geometry_programMarek Olšák2015-02-241-102/+30
| | | | | | | | | Mostly dead code or code that didn't do anything. Computing gs_num_outputs at the end was also useless. It's already set correctly. Reviewed-by: Dave Airlie <[email protected]>
* st/mesa: inline st_free_tokensMarek Olšák2015-02-243-17/+3
| | | | Reviewed-by: Dave Airlie <[email protected]>
* st/mesa: cleanup st_geometry_program structureMarek Olšák2015-02-243-74/+36
| | | | | | | It's full of unused variables and variables only used in st_translate_geometry_program. Reviewed-by: Dave Airlie <[email protected]>
* mesa: add a missing GS support check in GetActiveUniformBlockivMarek Olšák2015-02-241-0/+2
| | | | Reviewed-by: Ian Romanick <[email protected]>
* r600g: Implement GL_ARB_draw_indirect for EG/CMGlenn Kennard2015-02-2412-38/+133
| | | | | | | | | | | | | | | | Requires Evergreen/Cayman and radeon kernel module 2.41.0 or newer. Expected piglit fails due to hardware limitations: * arb_draw_indirect-draw-arrays-prim-restart Restarts not applied for DrawArrays commands * arb_draw_indirect-vertexid Base vertex offset is not included in vertex id Marek: bump vgt_state num_dw by 3 (= space needed for one register write) Signed-off-by: Glenn Kennard <[email protected]> Signed-off-by: Marek Olšák <[email protected]>
* freedreno/a4xx: aniso filteringRob Clark2015-02-241-4/+6
| | | | Signed-off-by: Rob Clark <[email protected]>
* freedreno: update generated headersRob Clark2015-02-245-5/+20
| | | | Signed-off-by: Rob Clark <[email protected]>
* freedreno/a4xx: add ARB_instanced_arrays supportRob Clark2015-02-242-5/+4
| | | | Signed-off-by: Rob Clark <[email protected]>
* freedreno/a4xx: handle index_bias (i.e. base_vertex)Rob Clark2015-02-241-1/+1
| | | | Signed-off-by: Rob Clark <[email protected]>
* freedreno/a4xx: add support for vertexid and instanceid sysvalsRob Clark2015-02-242-11/+24
| | | | | | ir3 bits of it already in place from a3xx patch.. Signed-off-by: Rob Clark <[email protected]>
* freedreno/a4xx: pass number of instances to drawRob Clark2015-02-243-6/+7
| | | | | | | a4xx has it's own draw packet, so needs equivalent update to what a3xx already got. Signed-off-by: Rob Clark <[email protected]>
* docs: add news item and link release notes for mesa 10.4.5Emil Velikov2015-02-242-0/+7
| | | | Signed-off-by: Emil Velikov <[email protected]>
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-20 15:55:21 +0000