2011-08-31 |
Eric Christopher | Rework this conditional a bit. |
blob | commitdiff | raw |
2011-08-31 |
Bruno Cardoso Lopes | - Move all MOVSS and MOVSD patterns close to their... |
blob | commitdiff | raw | diff to current |
2011-08-31 |
Bruno Cardoso Lopes | Teach more places to use VMOVAPS,VMOVUPS instead of... |
blob | commitdiff | raw | diff to current |
2011-08-31 |
Evan Cheng | Fix (movhps load) lowering / pattern to match more... |
blob | commitdiff | raw | diff to current |
2011-08-30 |
Rafael Espindola | Adds support for variable sized allocas. For a variable... |
blob | commitdiff | raw | diff to current |
2011-08-30 |
Rafael Espindola | Adds a SelectionDAG node X86SegAlloca which will be... |
blob | commitdiff | raw | diff to current |
2011-08-30 |
Rafael Espindola | Emit segmented-stack specific code into function prolog... |
blob | commitdiff | raw | diff to current |
2011-08-29 |
Eli Friedman | Explicitly zero out parts of a vector which are require... |
blob | commitdiff | raw | diff to current |
2011-08-27 |
Benjamin Kramer | Silence GCC warnings and make an array const. |
blob | commitdiff | raw | diff to current |
2011-08-26 |
Eli Friedman | Add support for generating CMPXCHG16B on x86-64 for... |
blob | commitdiff | raw | diff to current |
2011-08-25 |
Bruno Cardoso Lopes | Add support for AVX 256-bit version of MOVDDUP! |
blob | commitdiff | raw | diff to current |
2011-08-25 |
Bruno Cardoso Lopes | Make isMOVDDUP mask check more strict and update comments! |
blob | commitdiff | raw | diff to current |
2011-08-25 |
Bruno Cardoso Lopes | Add support for 256-bit versions of VSHUFPD and VSHUFPS. |
blob | commitdiff | raw | diff to current |
2011-08-24 |
Bruno Cardoso Lopes | Move all SHUFP* patterns close to the SHUFP* definition... |
blob | commitdiff | raw | diff to current |
2011-08-24 |
Eli Friedman | Hook up 64-bit atomic load/store on x86-32. I plan... |
blob | commitdiff | raw | diff to current |
2011-08-24 |
Eli Friedman | Fix whitespace. |
blob | commitdiff | raw | diff to current |
2011-08-24 |
Eli Friedman | Basic x86 code generation for atomic load and store... |
blob | commitdiff | raw | diff to current |
2011-08-24 |
Craig Topper | Break 256-bit vector int add/sub/mul into two 128-bit... |
blob | commitdiff | raw | diff to current |
2011-08-23 |
Bruno Cardoso Lopes | Fix a nasty bug where a v4i64 was being wrong emitted... |
blob | commitdiff | raw | diff to current |
2011-08-23 |
Nick Lewycky | PerformSubCombine to work on integers larger than i128... |
blob | commitdiff | raw | diff to current |
2011-08-23 |
Craig Topper | Add support for breaking 256-bit v16i16 and v32i8 VSETC... |
blob | commitdiff | raw | diff to current |
2011-08-22 |
Bruno Cardoso Lopes | Add support for breaking 256-bit int VETCC into two... |
blob | commitdiff | raw | diff to current |
2011-08-17 |
Bruno Cardoso Lopes | Fix PR10688. Add support for spliting 256-bit vector... |
blob | commitdiff | raw | diff to current |
2011-08-17 |
Bruno Cardoso Lopes | Introduce matching patterns for vbroadcast AVX instruct... |
blob | commitdiff | raw | diff to current |
2011-08-17 |
Bruno Cardoso Lopes | Update comments about vector splat handling in x86 |
blob | commitdiff | raw | diff to current |
2011-08-17 |
Bruno Cardoso Lopes | Now that we have a canonical way to handle 256-bit... |
blob | commitdiff | raw | diff to current |
2011-08-16 |
Bruno Cardoso Lopes | Instead of always leaving the work to the generic legal... |
blob | commitdiff | raw | diff to current |
2011-08-15 |
Bruno Cardoso Lopes | Fix PR10656. It's only profitable to use 128-bit insert... |
blob | commitdiff | raw | diff to current |
2011-08-12 |
Bruno Cardoso Lopes | Fix comment! |
blob | commitdiff | raw | diff to current |
2011-08-12 |
Bruno Cardoso Lopes | The VPERM2F128 is a AVX instruction which permutes... |
blob | commitdiff | raw | diff to current |
2011-08-11 |
Bruno Cardoso Lopes | Add a dag combine to xform 256-bit shuffles into simple... |
blob | commitdiff | raw | diff to current |
2011-08-11 |
Bruno Cardoso Lopes | Fix PR10492 by teaching MOVHLPS and MOVLPS mask matchin... |
blob | commitdiff | raw | diff to current |
2011-08-11 |
Nadav Rotem | Add a comment, per Bruno's CR. |
blob | commitdiff | raw | diff to current |
2011-08-11 |
Nadav Rotem | [AVX] If the data which is going to be saved is already... |
blob | commitdiff | raw | diff to current |
2011-08-11 |
Bruno Cardoso Lopes | Splats for v8i32/v8f32 can be handled by VPERMILPSY... |
blob | commitdiff | raw | diff to current |
2011-08-11 |
Bruno Cardoso Lopes | Use the splat index to generate the desired shuffle... |
blob | commitdiff | raw | diff to current |
2011-08-11 |
Eli Friedman | Fix X86TargetLowering::LowerExternalSymbol so that... |
blob | commitdiff | raw | diff to current |
2011-08-10 |
Nadav Rotem | When performing a truncating store, it is sometimes... |
blob | commitdiff | raw | diff to current |
2011-08-10 |
Bruno Cardoso Lopes | Fix a bug in vpermilps mask checking. Fix PR10560 |
blob | commitdiff | raw | diff to current |
2011-08-09 |
Bruno Cardoso Lopes | Add 256-bit support for v8i32, v4i64 and v4f64 ISD... |
blob | commitdiff | raw | diff to current |
2011-08-09 |
Bruno Cardoso Lopes | Use fp unpack instructions to unpack int types. Until... |
blob | commitdiff | raw | diff to current |
2011-08-09 |
Bruno Cardoso Lopes | Revert r137114 |
blob | commitdiff | raw | diff to current |
2011-08-09 |
Bruno Cardoso Lopes | Handle sitofp between v4f64 <- v4i32. Fix PR10559 |
blob | commitdiff | raw | diff to current |
2011-08-09 |
Bruno Cardoso Lopes | Make LowerVSETCC aware of AVX types and add patterns... |
blob | commitdiff | raw | diff to current |
2011-08-08 |
Bruno Cardoso Lopes | Add support for several vector shifts operations while... |
blob | commitdiff | raw | diff to current |
2011-08-04 |
Evan Cheng | Fix an obvious type. Patch by Ivan Krasin. |
blob | commitdiff | raw | diff to current |
2011-08-04 |
Bill Wendling | Only access both operands of an INSERT_SUBVECTOR if... |
blob | commitdiff | raw | diff to current |
2011-08-03 |
Benjamin Kramer | Remove unused variables. |
blob | commitdiff | raw | diff to current |
2011-08-02 |
Eli Friedman | Don't create a ridiculous EXTRACT_ELEMENT. PR10563. |
blob | commitdiff | raw | diff to current |
2011-08-02 |
Bruno Cardoso Lopes | Make this kind of lowering to be supported by 256-bit... |
blob | commitdiff | raw | diff to current |
2011-08-01 |
Bruno Cardoso Lopes | Add v4f64 -> v2f32 fp_round support. Also add a testcas... |
blob | commitdiff | raw | diff to current |
2011-08-01 |
Bruno Cardoso Lopes | Teach PreprocessISelDAG to be aware of vector types... |
blob | commitdiff | raw | diff to current |
2011-08-01 |
Bruno Cardoso Lopes | Lower CONCAT_VECTORS to use two VINSERTF128 instruction... |
blob | commitdiff | raw | diff to current |
2011-08-01 |
Bruno Cardoso Lopes | Since vectors with all ones can't be created with a... |
blob | commitdiff | raw | diff to current |
2011-07-29 |
Eli Friedman | Misc optimizer+codegen work for 'cmpxchg' and 'atomicrm... |
blob | commitdiff | raw | diff to current |
2011-07-29 |
Bruno Cardoso Lopes | Fix two tests that I crashed in the previous commits... |
blob | commitdiff | raw | diff to current |
2011-07-29 |
Bruno Cardoso Lopes | Match VPERMIL masks more strictly and update the target... |
blob | commitdiff | raw | diff to current |
2011-07-29 |
Bruno Cardoso Lopes | Add DecodeShuffle shuffle support for VPERMIPD variantes |
blob | commitdiff | raw | diff to current |
2011-07-29 |
Bruno Cardoso Lopes | Fix a bug while generating target specific VPERMIL... |
blob | commitdiff | raw | diff to current |
2011-07-29 |
Bruno Cardoso Lopes | Enable usage of SSE4 extracts and inserts in their... |
blob | commitdiff | raw | diff to current |
2011-07-29 |
Bruno Cardoso Lopes | Cleanup PALIGNR handling and remove the old palign... |
blob | commitdiff | raw | diff to current |
2011-07-28 |
Bruno Cardoso Lopes | Invert the subvector insertion to be more likely to... |
blob | commitdiff | raw | diff to current |
2011-07-28 |
Bruno Cardoso Lopes | Add SINT_TO_FP and FP_TO_SINT support for v8i32 types... |
blob | commitdiff | raw | diff to current |
2011-07-27 |
Eli Friedman | Code generation for 'fence' instruction. |
blob | commitdiff | raw | diff to current |
2011-07-27 |
Jeffrey Yasskin | Explicitly cast narrowing conversions inside {}s that... |
blob | commitdiff | raw | diff to current |
2011-07-27 |
Bruno Cardoso Lopes | Move some code around to open opportunity for more... |
blob | commitdiff | raw | diff to current |
2011-07-27 |
Bruno Cardoso Lopes | The vpermilps and vpermilpd have different behaviour... |
blob | commitdiff | raw | diff to current |
2011-07-26 |
Benjamin Kramer | Add a neat little two's complement hack for x86. |
blob | commitdiff | raw | diff to current |
2011-07-26 |
Bruno Cardoso Lopes | Recognize unpckh* masks and match 256-bit versions... |
blob | commitdiff | raw | diff to current |
2011-07-26 |
Eli Friedman | Prevent x86-specific DAGCombine from creating nodes... |
blob | commitdiff | raw | diff to current |
2011-07-26 |
Bruno Cardoso Lopes | More movsldup/movshdup cleanup. Rewrite the mask matchi... |
blob | commitdiff | raw | diff to current |
2011-07-26 |
Bruno Cardoso Lopes | More cleanup, subtarget info isn't used here. |
blob | commitdiff | raw | diff to current |
2011-07-25 |
Bruno Cardoso Lopes | Codegen allonesvector better while using AVX: vpcmpeqd... |
blob | commitdiff | raw | diff to current |
2011-07-25 |
Bruno Cardoso Lopes | - Handle special scalar_to_vector case: splats. Using... |
blob | commitdiff | raw | diff to current |
2011-07-25 |
Bruno Cardoso Lopes | Reintroduce r135730, this is indeed the right approach... |
blob | commitdiff | raw | diff to current |
2011-07-25 |
Eli Friedman | Get rid of an incorrect optimization for shuffles with... |
blob | commitdiff | raw | diff to current |
2011-07-22 |
Rafael Espindola | Turn shuffles into unpacks for VT == MVT::v2i64 and... |
blob | commitdiff | raw | diff to current |
2011-07-22 |
Dan Gohman | Fix x86's XALUO lowering to return its replacement... |
blob | commitdiff | raw | diff to current |
2011-07-22 |
Benjamin Kramer | GCC complains about the angle of this line. |
blob | commitdiff | raw | diff to current |
2011-07-22 |
Bruno Cardoso Lopes | Remove the 128-bit special handling from SCALAR_TO_VECT... |
blob | commitdiff | raw | diff to current |
2011-07-22 |
Bruno Cardoso Lopes | -Inspected a AVX code block added by someone in early... |
blob | commitdiff | raw | diff to current |
2011-07-22 |
Bruno Cardoso Lopes | Add a DAGCombine for transforming 128->256 casts into... |
blob | commitdiff | raw | diff to current |
2011-07-22 |
Bruno Cardoso Lopes | Introduce a new function to lower 256-bit vectors which... |
blob | commitdiff | raw | diff to current |
2011-07-22 |
Bruno Cardoso Lopes | Rename function to be more specific and be more strict... |
blob | commitdiff | raw | diff to current |
2011-07-21 |
Bruno Cardoso Lopes | - Register v16i16 as valid VR256 register class |
blob | commitdiff | raw | diff to current |
2011-07-21 |
Bruno Cardoso Lopes | Add support for 256-bit versions of VPERMIL instruction... |
blob | commitdiff | raw | diff to current |
2011-07-21 |
Bruno Cardoso Lopes | Improve splat promotion to handle AVX types: v32i8... |
blob | commitdiff | raw | diff to current |
2011-07-21 |
Bruno Cardoso Lopes | Tidy up code |
blob | commitdiff | raw | diff to current |
2011-07-20 |
Evan Cheng | Goodbye TargetAsmInfo. This eliminate last bit of CodeG... |
blob | commitdiff | raw | diff to current |
2011-07-18 |
Evan Cheng | Sink getDwarfRegNum, getLLVMRegNum, getSEHRegNum from... |
blob | commitdiff | raw | diff to current |
2011-07-18 |
Chris Lattner | land David Blaikie's patch to de-constify Type, with... |
blob | commitdiff | raw | diff to current |
2011-07-15 |
Bruno Cardoso Lopes | Fix a couple of things: |
blob | commitdiff | raw | diff to current |
2011-07-14 |
Eric Christopher | Check register class matching instead of width of type... |
blob | commitdiff | raw | diff to current |
2011-07-14 |
Nadav Rotem | [VECTOR-SELECT] |
blob | commitdiff | raw | diff to current |
2011-07-13 |
Bruno Cardoso Lopes | Make X86ISD::ANDNP more general and Codegen 256-bit... |
blob | commitdiff | raw | diff to current |
2011-07-13 |
Bruno Cardoso Lopes | The target specific node PANDN name is misleading.... |
blob | commitdiff | raw | diff to current |
2011-07-08 |
Julien Lerouge | Add _allrem, _aullrem and _allmul to the runtime for... |
blob | commitdiff | raw | diff to current |
2011-07-08 |
Cameron Zwarich | Add an intrinsic and codegen support for fused multiply... |
blob | commitdiff | raw | diff to current |
2011-07-08 |
Nick Lewycky | Let the inline asm 'q' constraint match float, and... |
blob | commitdiff | raw | diff to current |
2011-07-08 |
Eric Christopher | Go ahead and emit the barrier on x86-64 even without... |
blob | commitdiff | raw | diff to current |
next |