summaryrefslogtreecommitdiff
AgeCommit message (Expand)AuthorFilesLines
2021-04-02WIP: nir/algebraic: Add algebraic opt for fge comparison with identical opera...NaNu-NaNuIan Romanick1-0/+3
2021-04-02nir/algebraic: Use range analysis to "lower" some fsignIan Romanick1-0/+10
2021-04-02WIP: intel/fs: Allow SEL peephole when there is no else blockIan Romanick1-2/+41
2021-04-02intel/cfg: Document bblock_t::numIan Romanick1-0/+6
2021-04-02nir/algebraic: Detect open-coded bcselIan Romanick1-0/+9
2021-04-02WIP: nir/algebraic: Replace multiplication of b2f with a bcselIan Romanick1-0/+4
2021-04-02WIP: nir/algebraic: Add patterns for fpow(1.0, x) and fpow(x, 0.0)Ian Romanick1-0/+10
2021-04-02nir/algebraic: Remove redundant logical notIan Romanick1-0/+3
2021-04-02nir/algebraic: Replace multiplication of b2f with multiplication of minIan Romanick1-0/+23
2021-04-02nir/algebraic: Remove redundant multiply with b2fIan Romanick1-0/+4
2021-04-02nir/algebraic: Rearrange comparisons to encourage constant foldingIan Romanick1-0/+8
2021-04-02WIP: nir/algebraic: Simplify more open-coded fmin and fmaxIan Romanick1-0/+16
2021-04-02nir/algebraic: Duplicate patterns to recognize open-coded fmin / fmax in late...Ian Romanick1-0/+8
2021-04-02nir/algebraic: Simplify elaborate !(0 < a) comparisonIan Romanick1-0/+14
2021-04-02nir/algebraic: Simplify / generalize pattern to convert fmax to fsatIan Romanick1-5/+5
2021-04-02WIP: nir/algebraic: Simplify some more Inf and NaN avoidance codeIan Romanick1-0/+7
2021-04-02WIP: nir/algebraic: Add exact versions of some arithmetic identitiesIan Romanick1-0/+2
2021-04-02WIP: nir/algebraic: NaN foldingIan Romanick2-0/+36
2021-04-02nir/algebraic: Don't try to avoid NaN fmin / fmax resultsIan Romanick1-0/+22
2021-04-02WIP: nir/algebraic: Restore a pattern by requiring the operand be a numberIan Romanick1-0/+15
2021-04-02nir/range_analysis: Improve "is a number" range analysis for fexp2 and fpowIan Romanick1-6/+106
2021-04-02nir/range_analysis: Improve "is a number" range analysis for frcpIan Romanick1-6/+7
2021-04-02nir/range_analysis: Improve "is a number" range analysis for fsqrt and frsqIan Romanick1-1/+34
2021-04-02nir/range_analysis: Range tracking for flog2Ian Romanick1-0/+56
2021-04-02nir/range_analysis: Improve "is finite" range analysis for faddIan Romanick1-0/+24
2021-04-02nir/range_analysis: Improve range analysis for integer load_constIan Romanick1-0/+68
2021-04-02nir/range_analysis: Improve analysis when both fmul operands are integralIan Romanick1-4/+24
2021-04-02WIP: nir/range_analysis: Improve analysis of ffma and flrpIan Romanick1-152/+155
2021-04-02nir/algebraic: Eliminate a constant if both bcsel choices are constantsIan Romanick1-0/+12
2021-04-02nir/algebraic: Simplify some Inf and NaN avoidance codeIan Romanick1-0/+21
2021-04-02squash! WIP: nir/algebraic: Require sources of logic-joined comparisons be nu...Ian Romanick1-4/+0
2021-04-02WIP: nir/algebraic: Require sources of logic-joined comparisons be numbersIan Romanick1-4/+4
2021-04-02squash! squash! nir/algebraic: Mark a couple rules precise by requiring the o...Ian Romanick1-10/+0
2021-04-02squash! nir/algebraic: Mark a couple rules precise by requiring the operand b...Ian Romanick1-2/+0
2021-04-02nir/algebraic: Mark a couple rules precise by requiring the operand be a numberIan Romanick1-0/+2
2021-04-02nir/algebraic: Small optimizations for SpvOpFOrdNotEqual and SpvOpFUnordEqualIan Romanick1-0/+14
2021-04-02spirv: Generate shorter code for SpvOpFUnord comparisonsIan Romanick1-10/+23
2021-04-02spirv: SpvOpFUnordNotEqual doesn't need special treatmentIan Romanick1-1/+0
2021-04-02spirv: Minor cleanup in SpvOpFOrdNotEqualIan Romanick1-11/+1
2021-04-02spirv: Silence unused parameter warnings in vtn_alu.cIan Romanick1-4/+7
2021-04-02FYI: nir/algebraic: Eliminate useless fsat() on operand of comparison w/value...Ian Romanick1-0/+16
2021-04-02nir/algebraic: Add algebraic opt for float comparisons with identical operands.Ian Romanick1-3/+8
2021-04-02nir/algebraic: Rearrange some logic-joined comparisons and reduceIan Romanick1-0/+10
2021-04-02nir/algebraic: Mark some more logic-joined comparison reductions as exactIan Romanick1-4/+4
2021-04-02nir/algebraic: Mark some more comparison reductions exactIan Romanick1-4/+9
2021-04-02nir/algebraic: Equality comparison inversions require sources be numbersIan Romanick1-2/+2
2021-04-02nir/algebraic: Invert comparisons less oftenIan Romanick1-10/+10
2021-04-02nir/algebraic: Tautology replacements require sources be numbersIan Romanick2-6/+46
2021-04-02nir/algebraic: Remove some optimizations of comparisons with fsatIan Romanick1-29/+18
2021-04-02ci: Uprev piglit to 6a4be9e9946d ("piglit: NOTE! Default branch is now main")Eric Anholt14-80/+76
generated by cgit v1.2.3 (git 2.39.1) at 2024-05-16 00:08:37 +0000