Re-apply 68552.
[oota-llvm.git] / lib / Target / X86 / X86InstrSSE.td
2009-04-08 Rafael EspindolaRe-apply 68552.
2009-04-07 Bill WendlingTemporarily revert r68552. This was causing a failure...
2009-04-07 Rafael EspindolaReduce code duplication on the TLS implementation.
2009-02-26 Evan ChengADDS{D|S}rr_Int and MULS{D|S}rr_Int are not commutable...
2009-02-23 Nate BegemanGenerate better code for v8i16 shuffles on SSE2
2009-02-10 Evan ChengHandle llvm.x86.sse2.maskmov.dqu in 64-bit.
2009-02-05 Evan ChengA few more isAsCheapAsAMove.
2009-01-28 Evan ChengThe memory alignment requirement on some of the mov...
2009-01-09 Dan GohmanWhitespace and other minor adjustments to make SSE...
2008-12-18 Mon P WangFixed x86 code generation of multiple for v2i64. It...
2008-12-03 Dan GohmanRename isSimpleLoad to canFoldAsLoad, to better reflect...
2008-12-03 Dan GohmanMark x86's V_SET0 and V_SETALLONES with isSimpleLoad...
2008-10-17 Evan ChengFix lfence and mfence encoding. These look like MRM5r...
2008-10-16 Dan GohmanFix the predicate for memop64 to be a regular load...
2008-10-15 Dan GohmanNow that predicates can be composed, simplify several of
2008-10-10 Dale JohannesenFix SSE4.1 roundss, roundsd. While the instructions...
2008-10-07 Anders CarlssonCertain patterns involving the "movss" instruction...
2008-10-02 Bill Wendling"The original bug was a complaint that _mm_srli_si128...
2008-09-26 Evan ChengImplement "punpckldq %xmm0, $xmm0" as "pshufd $0x50...
2008-09-26 Evan Chengunpckhps requires sse1, punpckhdq requires sse2.
2008-09-25 Evan ChengWith sse3 and when the source is a load or has multiple...
2008-09-25 Evan Chengpmovsxbq etc. requires sse4.1.
2008-09-24 Evan ChengFix patterns for SSE4.1 move and sign extend instructio...
2008-09-12 Dan GohmanRename ConstantSDNode::getValue to getZExtValue, for...
2008-09-05 Eli FriedmanFix for PR2687: Add patterns to match sint_to_fp and...
2008-08-28 Evan ChengFsFLD0S{S|D} and V_SETALLONES are as cheap as moves.
2008-08-20 Dan GohmanTablegen generated code already tests the opcode value...
2008-08-08 Dan GohmanAdd an EXTRACTPSmr pattern to match the pattern that
2008-08-05 Evan ChengFix PR2620: Fix X86cmppd selection code so it expects...
2008-07-17 Nate BegemanFix a typo in last commit
2008-07-17 Nate BegemanSSE codegen for vsetcc nodes
2008-07-10 Evan ChengFix for PR2472. Use movss to set lower 32-bits of a...
2008-06-16 Evan ChengHorizontal-add instructions are not commutative.
2008-06-16 Evan Chengmpsadbw is commutable.
2008-06-13 Duncan SandsDisable some DAG combiner optimizations that may be
2008-05-29 Evan ChengImplement vector shift up / down and insert zero with...
2008-05-28 Dan GohmanFix the encoding for two more "rm" instructions that...
2008-05-28 Mon P WangFixed X86 encoding error CVTPS2PD and CVTPD2PS when...
2008-05-24 Evan ChengEliminate x86.sse2.punpckh.qdq and x86.sse2.punpckl...
2008-05-24 Evan ChengEliminate x86.sse2.movs.d, x86.sse2.shuf.pd, x86.sse2...
2008-05-24 Evan ChengRemove x86.sse2.loadh.pd and x86.sse2.loadl.pd. These...
2008-05-23 Evan ChengUse movlps / movhps to modify low / high half of 16...
2008-05-23 Evan ChengFix a duplicated pattern.
2008-05-23 Dan GohmanUse PMULDQ for v2i64 multiplies when SSE4.1 is availabl...
2008-05-23 Evan ChengBug: rcpps can only folds a load if the address is...
2008-05-22 Evan ChengAdd missing patterns.
2008-05-20 Evan Chengmovsd and movq do not require 16-byte alignment. This...
2008-05-13 Nate BegemanFix one more encoding bug.
2008-05-13 Nate BegemanFix and encoding error in the psrad xmm, imm8 instruction.
2008-05-12 Nate BegemanTeach Legalize how to scalarize VSETCC
2008-05-12 Nate BegemanInitial X86 codegen support for VSETCC.
2008-05-10 Evan ChengSome clean up.
2008-05-09 Evan ChengAdd a pattern to do move the low element of a v4f32...
2008-05-09 Evan ChengHandle a few more cases of folding load i64 into xmm...
2008-05-08 Evan ChengUse movq to move low half of XMM register and zero...
2008-05-08 Evan ChengHandle vector move / load which zero the destination...
2008-05-03 Evan ChengAdd separate intrinsics for MMX / SSE shifts with i32...
2008-05-02 Evan Cheng80 column violation.
2008-04-20 Chris LattnerA better fix for my previous patch, MOVZQI2PQIrr just...
2008-04-16 Dan GohmanAdd support for the form of the SSE41 extractps instruc...
2008-04-10 Chris LattnerFix the x86-64 side of PR2108 by adding a v2f64 version of
2008-04-05 Evan ChengFavors pshufd over shufps when shuffling elements from...
2008-03-26 Evan ChengFix some SSE4.1 instruction encoding bugs.
2008-03-24 Evan Cheng- SSE4.1 extractfps extracts a f32 into a gr32 register...
2008-03-16 Nate BegemanAdd a couple missing SSE4 instructions
2008-03-15 Evan ChengReplace all target specific implicit def instructions...
2008-03-14 Evan ChengFix some 80 col violations.
2008-03-14 Evan ChengFix a number of encoding bugs. SSE 4.1 instructions...
2008-03-12 Evan ChengClean up my own mess.
2008-03-08 Evan ChengImplement x86 support for @llvm.prefetch. It correspond...
2008-03-05 Evan ChengisTwoAddress = 1 -> Constraints.
2008-03-05 Evan ChengPSLLWri etc. are two-address instructions.
2008-02-18 Evan Cheng- When DAG combiner is folding a bit convert into a...
2008-02-16 Andrew Lenharthllvm.memory.barrier, and impl for x86 and alpha
2008-02-12 Nate BegemanSSE4.1 64b integer insert/extract pattern support
2008-02-11 Nate BegemanEnable SSE4 codegen and pattern matching.
2008-02-10 Nate Begemanxmm0 variable blends
2008-02-09 Nate Begemanmemopv16i8 had wrong alignment requirement, would have...
2008-02-09 Nate BegemanSkeleton of insert and extract matching, more to come
2008-02-04 Nate BegemanThe rest of the SSE4.1 intrinsic patterns that are...
2008-02-04 Nate BegemanSome more SSE 4.1 intrinsic patterns.
2008-02-03 Nate BegemanSSE 4.1 Intrinsics and detection
2008-01-24 Chris LattnerSignificantly simplify and improve handling of FP funct...
2008-01-11 Chris Lattneradd some missing flags.
2008-01-10 Chris LattnerStart inferring side effect information more aggressive...
2008-01-10 Chris Lattnerremove explicit sets of 'neverHasSideEffects' that...
2008-01-06 Chris Lattnerrename isLoad -> isSimpleLoad due to evan's desire...
2007-12-29 Chris LattnerRemove attribution from file headers, per discussion...
2007-12-20 Evan ChengFix JIT encoding for CMPSD as well.
2007-12-17 Bill WendlingAdd "mayHaveSideEffects" and "neverHasSideEffects"...
2007-12-16 Chris LattnerFix the JIT encoding of cmp*ss, which aborts with this...
2007-12-15 Evan ChengMake better use of instructions that clear high bits...
2007-12-12 Evan ChengImplicit def instructions, e.g. X86::IMPLICIT_DEF_GR32...
2007-12-06 Evan ChengRemove a bogus optimization. It's not possible to do...
2007-11-25 Chris LattnerFix a long standing deficiency in the X86 backend:...
2007-11-17 Nate BegemanAdd support for vectors to int <-> float casts.
2007-10-30 Dale JohannesenAdd missing SSE builtins: CVTPD2PI, CVTPS2PI,
2007-10-12 Arnold SchwaighoferCorrected many typing errors. And removed 'nest' parame...
2007-10-11 Dale JohannesenAdd missing argument to PALIGNR
2007-10-06 Evan ChengAdded DAG xforms. e.g.
next