oota-llvm.git
2006-04-07 Reid SpencerFor PR723:
2006-04-07 Reid SpencerFor PR723:
2006-04-07 Reid SpencerIgnore stuff built by "make dist".
2006-04-07 Evan ChengA saner workaround. I hope.
2006-04-07 Evan ChengTemporary workaround for a Mac OSX specific issue.
2006-04-07 Evan ChengAdded more shuffle tests
2006-04-07 Chris LattnerWe have an assertion that checks that we do not encode...
2006-04-07 Chris LattnerConstant fold extractelement(zero, x) -> zero
2006-04-07 Chris LattnerFix inlining of insert/extract element constantexprs
2006-04-07 Evan ChengAdd code to RemapOperand() to handle Instruction::Extra...
2006-04-06 Evan ChengA MOVPS2SSmr, i.e. _mm_store_ss, encoding bug.
2006-04-06 Evan ChengRemove int_x86_sse_loadh_ps and int_x86_sse_loadl_ps...
2006-04-06 Evan Cheng- movlp{s|d} and movhp{s|d} support.
2006-04-06 Evan ChengNew entries.
2006-04-06 Evan Cheng1. If both vector operands of a vector_shuffle are...
2006-04-06 Andrew LenharthThis may be overconservative, but it lets the new cfe...
2006-04-06 Chris LattnerAdd an item
2006-04-06 Chris LattnerMake sure to return the result in the right type.
2006-04-06 Andrew LenharthMove this to lib/Analysis.
2006-04-06 Chris LattnerMatch vpku[hw]um(x,x).
2006-04-06 Chris LattnerAdd testcases for vpku[hw]um(x,x)
2006-04-06 Reid SpencerAdd the variable llvmgccmajvers to the site.exp file...
2006-04-06 Reid SpencerAvoid building the runtime libraries if llvm-gcc versio...
2006-04-06 Chris LattnerAdd support for matching vmrg(x,x) patterns
2006-04-06 Chris LattnerAdd vmrg(x,x) tests
2006-04-06 Reid SpencerFor support of new GCC v4, obtain the full and major...
2006-04-06 Andrew Lenharthfix some linking problems with the new gcc
2006-04-06 Chris LattnerAdd tests for vmrg[hl]*
2006-04-06 Chris LattnerThese intrinsics are now lowered by the FE.
2006-04-06 Chris LattnerPattern match vmrg* instructions, which are now lowered...
2006-04-06 Chris LattnerInfer element types for shuffle masks
2006-04-06 Chris Lattnerrename a method, to avoid confusion with llvm intrinsics.
2006-04-06 Chris Lattnertest vperm promotion
2006-04-06 Chris Lattnerremove two done items
2006-04-06 Chris LattnerLower vperm(x,y, mask) -> shuffle(x,y,mask) if mask...
2006-04-06 Chris Lattnerthis is no longer an intrinsic.
2006-04-06 Chris LattnerSupport pattern matching vsldoi(x,y) and vsldoi(x,x...
2006-04-06 Chris Lattnernew testcase
2006-04-06 Chris LattnerCompile the vpkuhum/vpkuwum intrinsics into vpkuhum...
2006-04-06 Chris LattnerAdd support for building the LLVM libraries and tools...
2006-04-06 Evan ChengPOR encoded as PAND, yikes.
2006-04-05 Evan ChengAn entry about comi / ucomi intrinsics.
2006-04-05 Evan ChengSupport for comi / ucomi intrinsics.
2006-04-05 Evan ChengAdded comi and ucomi SSE intrinsics.
2006-04-05 Chris LattnerAdd all of the data stream intrinsics and instructions...
2006-04-05 Chris Lattneradd altivec ds* intrinsics
2006-04-05 Chris LattnerFix a typo
2006-04-05 Chris LattnerFix CodeGen/PowerPC/2006-04-05-splat-ish.ll
2006-04-05 Chris Lattnernew testcase that was miscompiled to vspltisb
2006-04-05 Evan ChengHandle canonical form of e.g.
2006-04-05 Chris LattnerMake a vector live across blocks have the correct Vec...
2006-04-05 Chris Lattnernew testcase
2006-04-05 Evan ChengBogus assert
2006-04-05 Evan ChengFallthrough to expand if a VECTOR_SHUFFLE cannot be...
2006-04-05 Evan ChengExapnd a VECTOR_SHUFFLE to a BUILD_VECTOR if target...
2006-04-05 Jeff CohenFix more tablegen depedency issues in Visual Studio.
2006-04-05 Evan ChengSeparate out to 2 test cases
2006-04-05 Andrew Lenharthmake this test less exacting
2006-04-05 Andrew Lenharthrevert this, this is safe, if conservative. leave...
2006-04-05 Evan ChengHandle v8i16 shuffle that must be broken into a pair...
2006-04-05 Evan ChengAdd a new shuffle test case that requires pshuflw ...
2006-04-05 Chris Lattneradd vsl
2006-04-05 Chris LattnerGet the types right, third time is the charm. Add...
2006-04-05 Chris Lattneradd vmladduhm
2006-04-05 Chris Lattnercorrect the type of two intrinsics, add int_ppc_altivec...
2006-04-05 Chris LattnerAdd m[tf]vscr instructions.
2006-04-05 Chris LattnerAdd m[tf]vscr intrinsics.
2006-04-04 Chris Lattneradd a note
2006-04-04 Chris LattnerAdd missing byte merges.
2006-04-04 Chris LattnerAdd FP -> Int Conversions
2006-04-04 Chris Lattneradd average intrinsics
2006-04-04 Chris Lattneradd average intrinsics.
2006-04-04 Chris Lattneradd a note
2006-04-04 Chris LattnerFix some broken logic that would cause us to codegen...
2006-04-04 Evan ChengAdded pslldq and psrldq.
2006-04-04 Evan ChengAdded intrinsics to match __builtin_ia32_pslldqi128 and
2006-04-04 Evan ChengMinor fixes + naming changes.
2006-04-04 Chris LattnerHow could this ever have worked?
2006-04-04 Evan ChengPSHUF* encoding bugs.
2006-04-04 Chris LattnerMake sure to consider alignment of variable sized objects.
2006-04-04 Chris LattnerDo not create ZEXTLOAD's unless we are before legalize...
2006-04-04 Chris LattnerNew testcase
2006-04-04 Chris LattnerAsk legalize to promote all vector shuffles to be v16i8...
2006-04-04 Chris Lattner* Add supprot for SCALAR_TO_VECTOR operations where...
2006-04-04 Chris LattnerMove isShuffleLegal from TLI to Legalize.
2006-04-04 Chris Lattnernew testcase
2006-04-04 Chris LattnerSigned shr by a constant is not the same as sdiv by 2^k
2006-04-04 Evan Chengcmpps / cmppd encoding bug
2006-04-04 Chris LattnerFix the types for these intrinsics.
2006-04-04 Chris LattnerConstant fold bitconvert(undef)
2006-04-04 Chris LattnerAllow targets to have fine grained control over which...
2006-04-04 Evan ChengCompact some intrinsic definitions.
2006-04-04 Chris LattnerPlug in the byte and short splats
2006-04-03 Chris LattnerRevert accidentally committed hunks.
2006-04-03 Chris LattnerMake sure to mark unsupported SCALAR_TO_VECTOR operatio...
2006-04-03 Evan ChengSome SSE1 intrinsics: min, max, sqrt, etc.
2006-04-03 Chris Lattnerrevert previous patch
2006-04-03 Evan ChengUse movlpd to: store lower f64 extracted from v2f64.
2006-04-03 Chris LattnerForce use of a frame-pointer if there is anything on...
2006-04-03 Chris LattnerThe stack alignment is now computed dynamically, just...
next