Implement vector shift up / down and insert zero with ps{rl}lq / ps{rl}ldq.
[oota-llvm.git] / lib / Target / X86 / X86InstrSSE.td
2007-03-21 Evan ChengMark re-materializable instructions.
2007-03-04 Chris Lattneradd missing braces
2007-02-20 Evan ChengHow the heck did I forget patterns for llvm.x86.sse2...
2007-01-05 Evan Cheng- FCOPYSIGN custom lowering bug. Clear the sign bit...
2007-01-05 Evan ChengWith SSE2, expand FCOPYSIGN to a series of SSE bitwise...
2006-12-14 Evan Cheng- Rename MOVDSS2DIrr to MOVSS2DIrr for consistency...
2006-12-05 Chris LattnerIf we have ScalarSSE, we can select bitconvert into...
2006-11-16 Evan ChengCorrect instructions for moving data between GR64 and...
2006-11-10 Evan ChengDon't dag combine floating point select to max and...
2006-11-07 Evan ChengFixed a bug which causes x86 be to incorrectly match
2006-11-03 Chris Lattnerremove dead/redundant vars
2006-11-01 Evan ChengFix ldmxcsr JIT encoding.
2006-10-27 Evan ChengFixed a significant bug where unpcklpd is incorrectly...
2006-10-25 Evan ChengX86ISD::PEXTRW 3rd operand type is always target pointe...
2006-10-11 Evan ChengComplexPatterns sse_load_f32 and sse_load_f64 returns...
2006-10-09 Evan ChengDon't go too crazy with these AddComplexity. Try matchi...
2006-10-09 Evan ChengReflects ISD::LOAD / ISD::LOADX / LoadSDNode changes.
2006-10-07 Chris Lattnercompletely disable folding of loads into scalar sse...
2006-10-07 Chris Lattnerconvert packed FP add/sub/mul/div to use a multiclass.
2006-10-07 Chris Lattnerone multiclass now defines all 8 variants of binary...
2006-10-07 Chris LattnerSwitch ADD/MUL/DIV/SUB scalarsse fp ops to a multiclass
2006-10-07 Chris LattnerRandom acts of shrinkage
2006-10-07 Chris LattnerConvert pand/por/pxor to use multiclass
2006-10-07 Chris LattnerConvert some more instructions over to use a new multic...
2006-10-07 Chris LattnerFix a bug where PADDQrm printed paddd instead of paddq.
2006-10-07 Chris LattnerAdd multiclass for SSE2 instructions that correspond...
2006-10-07 Chris Lattnerrename:
2006-10-07 Chris LattnerConvert saturating PADD/PSUB's to use a multiclass
2006-10-07 Chris LattnerConvert PAVG*, PMADDWD, and PMUL* to use multiclasses.
2006-10-07 Chris LattnerFix typo in packsswb instr definition, where the load...
2006-10-07 Chris Lattnerhandle pmin/pmax with multiclasses
2006-10-07 Chris Lattnersimplify pack and shift intrinsics with multiclasses
2006-10-07 Chris LattnerUse a multiclass to simplify 'SSE2 Integer comparison'
2006-10-07 Chris Lattnermove class defns close to uses to make it easier to...
2006-10-07 Chris Lattnersimplify horizontal op definitions
2006-10-07 Chris Lattnerremove more unneeded type info
2006-10-07 Chris Lattnerremove unneeded definitions and type info
2006-10-07 Chris Lattnerremove some unneeded type info
2006-10-07 Chris Lattnersimplify patterns by merging in operand info
2006-10-07 Chris LattnerFactor operands into packed unary classes
2006-10-07 Chris Lattnerremove dead/duplicate instructions
2006-10-07 Chris LattnerPull operand info up into parent class for scalar sse...
2006-10-07 Chris Lattnerconvert the sole sd unary intrinsic to a multiclass...
2006-10-07 Chris Lattnerpull operand string into the multiclass
2006-10-07 Chris LattnerRemove RSQRTSS[rm] RCPSS[rm], which are dead.
2006-10-07 Chris Lattnereliminate redundancy
2006-10-03 Evan ChengThese don't have immediate operands.
2006-09-11 Evan ChengX86ISD::CMP now produces a chain as well as a flag...
2006-09-05 Evan ChengJIT encoding bug.
2006-07-25 Evan ChengCan't commute shufps. The high / low parts elements...
2006-07-07 Evan ChengX86 target specific DAG combine: turn build_vector...
2006-06-29 Evan ChengShould just use xorps to clear XMM registers for all...
2006-06-29 Evan ChengAlways use xorps to clear XMM registers.
2006-06-20 Chris LattnerRemove some ugly now-redundant casts.
2006-06-20 Chris LattnerFix some mismatched type constraints
2006-06-19 Evan ChengMinor clean up.
2006-06-15 Evan ChengType of vector extract / insert index operand should...
2006-05-31 Evan ChengRename instructions for consistency sake.
2006-05-31 Evan ChengSelect vector_shuffle v1, undef <2, 3, ?, ?> to MOVHLPS.
2006-05-30 Evan ChengMAXP{D|S} and MINP{D|S} are commutable.
2006-05-30 Evan ChengCommute shufps / shufpd.
2006-05-30 Evan ChengAllow shufps x, x, mask to be converted to pshufd x...
2006-05-16 Evan ChengX86 integer register classes naming changes. Make them...
2006-05-05 Chris LattnerTeach the code generator to use cvtss2sd as extload...
2006-05-03 Evan ChengUse movsd to shuffle in the lowest two elements of...
2006-04-25 Evan ChengFix a typo.
2006-04-25 Evan ChengExplicitly specify result type for def : Pat<> patterns...
2006-04-24 Evan ChengAdded X86 SSE2 intrinsics which can be represented...
2006-04-24 Evan ChengSome missing movlps, movhps, movlpd, and movhpd patterns.
2006-04-21 Evan ChengNow generating perfect (I think) code for "vector set...
2006-04-19 Evan ChengPrefer {p}unpack* and mov*dup over {p}shuf* as well.
2006-04-19 Evan Cheng- Renamed AddedCost to AddedComplexity.
2006-04-19 Evan ChengMore mov{h|l}p{d|s} patterns.
2006-04-19 Evan Cheng- More mov{h|l}ps patterns.
2006-04-18 Evan Cheng- PEXTRW cannot take a memory location as its first...
2006-04-18 Evan ChengName change for clarity sake
2006-04-18 Evan ChengName change for clarity sake
2006-04-18 Evan ChengLeft a pattern out
2006-04-18 Evan ChengFixed an encoding bug: movd from XMM to R32.
2006-04-17 Evan ChengUse movss to insert_vector_elt(v, s, 0).
2006-04-17 Evan ChengEncoding bug
2006-04-17 Evan ChengErrors in patterns preventing load folding
2006-04-16 Evan Chengmovduprm, movshduprm bugs
2006-04-16 Evan ChengEncoding bugs
2006-04-15 Evan ChengMore encoding bugs
2006-04-15 Evan Chengpslldrm, psrawrm, etc. encoding bug
2006-04-15 Evan Chenghsubp{s|d} encoding bug
2006-04-15 Evan ChengSilly bug
2006-04-14 Evan ChengSome clean up
2006-04-14 Evan ChengLast few SSE3 intrinsics.
2006-04-14 Evan ChengMisc. SSE2 intrinsics: clflush, lfench, mfence
2006-04-14 Evan Chengpcmpeq* and pcmpgt* intrinsics.
2006-04-14 Evan Chengpsll*, psrl*, and psra* intrinsics.
2006-04-13 Evan ChengDoh. PANDrm, etc. are not commutable.
2006-04-13 Evan Chengpsad, pmax, pmin intrinsics.
2006-04-13 Evan ChengVarious SSE2 packed integer intrinsics: pmulhuw, pavgw...
2006-04-13 Evan Chengpadds{b|w}, paddus{b|w}, psubs{b|w}, psubus{b|w} intrin...
2006-04-13 Evan ChengNaming inconsistency.
2006-04-12 Evan ChengSSE / SSE2 conversion intrinsics.
2006-04-12 Evan ChengAll "integer" logical ops (pand, por, pxor) are now...
next