| Commit message (Expand) | Author | Age | Files | Lines |
* | Fix PR1975: dag isel emitter produces patterns that isel wrong flag result. | Evan Cheng | 2008-02-05 | 1 | -0/+12 |
* | Add target triples to these so they don't fail on linux. | Chris Lattner | 2008-01-29 | 1 | -0/+2 |
* | Implement some dag combines that allow doing fneg/fabs/fcopysign in integer | Chris Lattner | 2008-01-27 | 1 | -0/+21 |
* | New test to verify that "merging 4 loads into a vec load" continues to work and | Chris Lattner | 2008-01-26 | 1 | -0/+35 |
* | Infer alignment of loads and increase their alignment when we can tell they are | Chris Lattner | 2008-01-26 | 1 | -0/+22 |
* | remove a useless xfailed test. | Chris Lattner | 2008-01-26 | 1 | -27/+0 |
* | If there's no instructions being emitted on X86 for a function, emit a | Bill Wendling | 2008-01-26 | 1 | -0/+6 |
* | Add target-specific dag combines for FAND(x,0) and FOR(x,0). This allows | Chris Lattner | 2008-01-25 | 1 | -0/+14 |
* | Significantly simplify and improve handling of FP function results on x86-32. | Chris Lattner | 2008-01-24 | 3 | -1/+29 |
* | take these with a pr # | Chris Lattner | 2008-01-24 | 2 | -2/+2 |
* | Let each target decide byval alignment. For X86, it's 4-byte unless the aggre... | Evan Cheng | 2008-01-23 | 4 | -4/+34 |
* | SSE varargs arguments are passed in memory. | Evan Cheng | 2008-01-22 | 1 | -0/+9 |
* | Test is correct again for the moment. | Dale Johannesen | 2008-01-18 | 1 | -3/+0 |
* | This commit changes: | Chris Lattner | 2008-01-17 | 1 | -0/+8 |
* | When a live virtual register is being clobbered by an implicit def, it is spi... | Evan Cheng | 2008-01-17 | 1 | -0/+35 |
* | Fixes a nasty dag combiner bug that causes a bunch of tests to fail at -O0. | Evan Cheng | 2008-01-16 | 1 | -0/+30 |
* | Trampoline support for x86-64. This looks like | Duncan Sands | 2008-01-16 | 1 | -0/+14 |
* | make sure to use a cpu that has sse. | Chris Lattner | 2008-01-16 | 1 | -1/+1 |
* | My previous commit had an incomplete message, it should have been: | Chris Lattner | 2008-01-16 | 1 | -0/+12 |
* | verify x86 generates ud2 for llvm.trap | Chris Lattner | 2008-01-15 | 1 | -0/+9 |
* | Disable for now. | Dale Johannesen | 2008-01-11 | 1 | -0/+3 |
* | Output sinl for a long double FSIN node, not sin. | Duncan Sands | 2008-01-10 | 1 | -0/+11 |
* | Codegen improvement has reduced one spill. | Evan Cheng | 2008-01-10 | 1 | -1/+1 |
* | Special copy SUnit's do not have SDNode's. | Evan Cheng | 2008-01-09 | 1 | -0/+32 |
* | Fix sse2.psrl.w and sse2.psrl.q definitions. | Evan Cheng | 2008-01-09 | 2 | -2/+36 |
* | Make load->store deletion a bit smarter. This allows us to compile this: | Chris Lattner | 2008-01-08 | 1 | -0/+11 |
* | Crashes llc when using Chris's new legalization logic. | Duncan Sands | 2008-01-08 | 1 | -0/+17 |
* | Update test to catch recent x86 insert regression and improvements | Nate Begeman | 2008-01-07 | 1 | -0/+3 |
* | fix this to use a valid triple. | Chris Lattner | 2008-01-02 | 1 | -1/+1 |
* | verify that aligned common support doesn't break. | Chris Lattner | 2008-01-02 | 1 | -0/+8 |
* | One readme entry is done, one is really easy (Evan, want to investigate | Chris Lattner | 2007-12-29 | 1 | -0/+10 |
* | upgrade this test | Chris Lattner | 2007-12-29 | 1 | -8/+9 |
* | Fold comparisons against a constant nan, and optimize ORD/UNORD | Chris Lattner | 2007-12-29 | 1 | -0/+11 |
* | this xform is implemented. | Chris Lattner | 2007-12-29 | 1 | -0/+9 |
* | Codegen: | Chris Lattner | 2007-12-29 | 1 | -0/+11 |
* | avoid going through a stack slot to convert from fpstack to xmm reg | Chris Lattner | 2007-12-29 | 1 | -0/+15 |
* | one fewer uncond branch with my codegenprepare hack for single-mbb backedges. | Chris Lattner | 2007-12-26 | 1 | -1/+1 |
* | Fix PR1872: SrcValue and SrcValueOffset should not be used to compute load / ... | Evan Cheng | 2007-12-18 | 1 | -0/+28 |
* | FIX for PR1799: When a load is unfolded from an instruction, check if it is a... | Evan Cheng | 2007-12-18 | 1 | -0/+35 |
* | Make better use of instructions that clear high bits; fix various 2-wide shuf... | Evan Cheng | 2007-12-15 | 4 | -2/+126 |
* | Fix ctlz and cttz. llvm definition requires them to return number of bits in ... | Evan Cheng | 2007-12-14 | 1 | -1/+11 |
* | Implement ctlz and cttz with bsr and bsf. | Evan Cheng | 2007-12-14 | 1 | -0/+16 |
* | Be extra careful with extension use optimation. Now turned on by default. | Evan Cheng | 2007-12-13 | 1 | -1/+1 |
* | Fold some and + shift in x86 addressing mode. | Evan Cheng | 2007-12-13 | 1 | -0/+21 |
* | Implicit def instructions, e.g. X86::IMPLICIT_DEF_GR32, are always re-materia... | Evan Cheng | 2007-12-12 | 1 | -0/+680 |
* | Allow vector integer constants to be created with | Dan Gohman | 2007-12-12 | 1 | -0/+18 |
* | Use shuffles to implement insert_vector_elt for i32, i64, f32, and f64. | Evan Cheng | 2007-12-12 | 4 | -28/+40 |
* | Add a test case for -optimize-ext-uses. | Evan Cheng | 2007-12-12 | 1 | -0/+19 |
* | Lower a build_vector with all constants into a constpool load unless it can b... | Evan Cheng | 2007-12-12 | 1 | -3/+10 |
* | - Improved v8i16 shuffle lowering. It now uses pshuflw and pshufhw as much as | Evan Cheng | 2007-12-11 | 2 | -24/+36 |