| Commit message (Expand) | Author | Age | Files | Lines |
* | Codegen allonesvector better while using AVX: vpcmpeqd + vinsertf128 | Bruno Cardoso Lopes | 2011-07-25 | 1 | -9/+44 |
* | - Handle special scalar_to_vector case: splats. Using a native 128-bit | Bruno Cardoso Lopes | 2011-07-25 | 1 | -1/+39 |
* | Reintroduce r135730, this is indeed the right approach, there is no | Bruno Cardoso Lopes | 2011-07-25 | 1 | -0/+18 |
* | Get rid of an incorrect optimization for shuffles with PALIGNR and simplify i... | Eli Friedman | 2011-07-25 | 1 | -15/+5 |
* | Turn shuffles into unpacks for VT == MVT::v2i64 and MVT::v2f64 | Rafael Espindola | 2011-07-22 | 1 | -9/+5 |
* | Fix x86's XALUO lowering to return its replacement values instead | Dan Gohman | 2011-07-22 | 1 | -4/+2 |
* | GCC complains about the angle of this line. | Benjamin Kramer | 2011-07-22 | 1 | -1/+1 |
* | Remove the 128-bit special handling from SCALAR_TO_VECTOR. This isn't | Bruno Cardoso Lopes | 2011-07-22 | 1 | -18/+0 |
* | -Inspected a AVX code block added by someone in early Feb. This was never used | Bruno Cardoso Lopes | 2011-07-22 | 1 | -74/+25 |
* | Add a DAGCombine for transforming 128->256 casts into a simple | Bruno Cardoso Lopes | 2011-07-22 | 1 | -7/+61 |
* | Introduce a new function to lower 256-bit vectors which are not | Bruno Cardoso Lopes | 2011-07-22 | 1 | -0/+14 |
* | Rename function to be more specific and be more strict about its usage | Bruno Cardoso Lopes | 2011-07-22 | 1 | -6/+9 |
* | - Register v16i16 as valid VR256 register class | Bruno Cardoso Lopes | 2011-07-21 | 1 | -5/+6 |
* | Add support for 256-bit versions of VPERMIL instruction. This is a new | Bruno Cardoso Lopes | 2011-07-21 | 1 | -0/+63 |
* | Improve splat promotion to handle AVX types: v32i8 and v16i16. Also | Bruno Cardoso Lopes | 2011-07-21 | 1 | -24/+87 |
* | Tidy up code | Bruno Cardoso Lopes | 2011-07-21 | 1 | -15/+5 |
* | Goodbye TargetAsmInfo. This eliminate last bit of CodeGen and Target in llvm-mc. | Evan Cheng | 2011-07-20 | 1 | -5/+2 |
* | Sink getDwarfRegNum, getLLVMRegNum, getSEHRegNum from TargetRegisterInfo down | Evan Cheng | 2011-07-18 | 1 | -3/+3 |
* | land David Blaikie's patch to de-constify Type, with a few tweaks. | Chris Lattner | 2011-07-18 | 1 | -15/+15 |
* | Fix a couple of things: | Bruno Cardoso Lopes | 2011-07-15 | 1 | -61/+52 |
* | Check register class matching instead of width of type matching | Eric Christopher | 2011-07-14 | 1 | -4/+4 |
* | [VECTOR-SELECT] | Nadav Rotem | 2011-07-14 | 1 | -2/+64 |
* | Make X86ISD::ANDNP more general and Codegen 256-bit VANDNP. A more | Bruno Cardoso Lopes | 2011-07-13 | 1 | -3/+5 |
* | The target specific node PANDN name is misleading. That happens because | Bruno Cardoso Lopes | 2011-07-13 | 1 | -7/+7 |
* | Add _allrem, _aullrem and _allmul to the runtime for MSVC. | Julien Lerouge | 2011-07-08 | 1 | -0/+6 |
* | Add an intrinsic and codegen support for fused multiply-accumulate. The intent | Cameron Zwarich | 2011-07-08 | 1 | -0/+6 |
* | Let the inline asm 'q' constraint match float, and on 64-bit double too. | Nick Lewycky | 2011-07-08 | 1 | -3/+3 |
* | Go ahead and emit the barrier on x86-64 even without sse2. The | Eric Christopher | 2011-07-08 | 1 | -3/+4 |
* | Add support for the X86 'l' constraint. | Eric Christopher | 2011-07-07 | 1 | -0/+1 |
* | Use getRegForInlineAsmConstraint instead of custom defining regclasses | Eric Christopher | 2011-06-29 | 1 | -52/+24 |
* | Clean up the handling of the x87 fp stack to make it more robust. | Jakob Stoklund Olesen | 2011-06-28 | 1 | -8/+3 |
* | Replace dyn_cast<> with cast<> since the cast is already guarded by the neces... | Chad Rosier | 2011-06-25 | 1 | -1/+1 |
* | Enable tail call optimization in the presence of a byval (x86-32 and x86-64). | Chad Rosier | 2011-06-25 | 1 | -0/+4 |
* | Hoist simple check above more complex checking to avoid unnecessary | Chad Rosier | 2011-06-24 | 1 | -5/+5 |
* | Remove TargetOptions.h dependency from X86Subtarget. | Evan Cheng | 2011-06-23 | 1 | -2/+25 |
* | Remove unused but set variables. | Benjamin Kramer | 2011-06-18 | 1 | -3/+4 |
* | Add a new function attribute, nonlazybind, which inhibits lazy-loading | John McCall | 2011-06-15 | 1 | -0/+20 |
* | Add a parameter to CCState so that it can access the MachineFunction. | Eric Christopher | 2011-06-08 | 1 | -19/+20 |
* | Followup to 132458, omit unnecessary stack copy when x87 input is a | Stuart Hastings | 2011-06-06 | 1 | -5/+22 |
* | Reapply 132424 with fixes. This fixes PR10068. | Stuart Hastings | 2011-06-03 | 1 | -0/+88 |
* | Have LowerOperandForConstraint handle multiple character constraints. | Eric Christopher | 2011-06-02 | 1 | -2/+6 |
* | Revert 132424 to fix PR10068. | Rafael Espindola | 2011-06-02 | 1 | -82/+0 |
* | Omit unnecessary stack copy when x87 input is a load. | Stuart Hastings | 2011-06-02 | 1 | -6/+17 |
* | Recommit 132404 with fixes. rdar://problem/5993888 | Stuart Hastings | 2011-06-01 | 1 | -0/+82 |
* | Revert 132404 to appease a buildbot. rdar://problem/5993888 | Stuart Hastings | 2011-06-01 | 1 | -82/+0 |
* | Add support for x86 CMPEQSS and friends. These instructions do a | Stuart Hastings | 2011-06-01 | 1 | -0/+82 |
* | FGETSIGN support for x86, using movmskps/pd. Will be enabled with a | Stuart Hastings | 2011-06-01 | 1 | -0/+16 |
* | Reverting 132105: it broke some LLVM-GCC DejaGNU tests. | Stuart Hastings | 2011-05-26 | 1 | -72/+13 |
* | Correctly handle a one-word struct passed byval on x86_64. | Stuart Hastings | 2011-05-26 | 1 | -13/+72 |
* | - Teach SelectionDAG::isKnownNeverZero to return true (op x, c) when c is | Evan Cheng | 2011-05-24 | 1 | -7/+17 |