Commits · fb4161ae88fd6071bb615a3b69e207ec2fca9f64 · Roger Ferrer / llvm-epi · GitLab

Aug 18, 2011
- Cleanup vector logical ops in AVX and add use int versions for simple · 3c7d6eb6
  Bruno Cardoso Lopes authored Aug 18, 2011
  3c7d6eb6
- Fix PR10688. Add support for spliting 256-bit vector shifts when the · 1a87fcb9
  Bruno Cardoso Lopes authored Aug 17, 2011
  1a87fcb9
Aug 17, 2011
- Allow the MCDisassembler to return a "soft fail" status code, indicating an... · a4043c4b
  Owen Anderson authored Aug 17, 2011
  a4043c4b
- Introduce matching patterns for vbroadcast AVX instruction. The idea is to · be5e9873
  Bruno Cardoso Lopes authored Aug 17, 2011
  be5e9873
- Update comments about vector splat handling in x86 · 6d33c7f3
  Bruno Cardoso Lopes authored Aug 17, 2011
  6d33c7f3
- Now that we have a canonical way to handle 256-bit splats: · ed786a34
  Bruno Cardoso Lopes authored Aug 17, 2011
  ed786a34
Aug 16, 2011
- Instead of always leaving the work to the generic legalizer when · 2e99f1b3
  Bruno Cardoso Lopes authored Aug 16, 2011
  2e99f1b3
- While I'm here, remove the "_alt" hacks to a series of INSERT_SUBREG and · c1676e41
  Bruno Cardoso Lopes authored Aug 15, 2011
  c1676e41
- Reorder declarations of vmovmskp* and also put the necessary AVX · 67005029
  Bruno Cardoso Lopes authored Aug 15, 2011
  67005029
- MCTargetAsmParser target match predicate support. · 120a96a7
  Jim Grosbach authored Aug 15, 2011
  120a96a7
Aug 15, 2011
- Fix PR10656. It's only profitable to use 128-bit inserts and extracts · cbe7feea
  Bruno Cardoso Lopes authored Aug 15, 2011
  cbe7feea
Aug 12, 2011
- Fix comment! · c53dd2ac
  Bruno Cardoso Lopes authored Aug 12, 2011
  c53dd2ac
- The VPERM2F128 is a AVX instruction which permutes between two 256-bit · f15dfe58
  Bruno Cardoso Lopes authored Aug 12, 2011
  f15dfe58
- Move code around and add comments · 960c8f71
  Bruno Cardoso Lopes authored Aug 12, 2011
  960c8f71
- Silence a bunch (but not all) "variable written but not read" warnings · a41634e3
  Duncan Sands authored Aug 12, 2011
  a41634e3
- findDeadCallerSavedReg fix: Missing NULL terminator in register arrays. · 210bf835
  Andrew Trick authored Aug 12, 2011
  210bf835
Aug 11, 2011
- Add a dag combine to xform 256-bit shuffles into simple vector · 8fbf023c
  Bruno Cardoso Lopes authored Aug 11, 2011
  8fbf023c
- Fix PR10492 by teaching MOVHLPS and MOVLPS mask matching to be more strict. · 043c8208
  Bruno Cardoso Lopes authored Aug 11, 2011
  043c8208
- Add a comment, per Bruno's CR. · efdd183f
  Nadav Rotem authored Aug 11, 2011
  efdd183f
- [AVX] If the data which is going to be saved is already in two XMM registers · 1542d5a0
  Nadav Rotem authored Aug 11, 2011
  1542d5a0
- Cleanup: Remove Int_ CVTSS2SI* forms · dbd1352c
  Bruno Cardoso Lopes authored Aug 11, 2011
  dbd1352c
- Splats for v8i32/v8f32 can be handled by VPERMILPSY. This was causing · a2d8bb97
  Bruno Cardoso Lopes authored Aug 11, 2011
  a2d8bb97
- Use the splat index to generate the desired shuffle. Otherwise we · 572c9aaf
  Bruno Cardoso Lopes authored Aug 11, 2011
  572c9aaf
- Fix X86TargetLowering::LowerExternalSymbol so that it actually works in... · 3ae39f8a
  Eli Friedman authored Aug 11, 2011
  3ae39f8a
Aug 10, 2011
- When performing a truncating store, it is sometimes possible to rearrange the · 410a11fe
  Nadav Rotem authored Aug 10, 2011
  410a11fe
- The following X86 pattern is incorrect: · 3ff111c1
  Bruno Cardoso Lopes authored Aug 10, 2011
  3ff111c1
- Fix a bug in vpermilps mask checking. Fix PR10560 · 278ffd7d
  Bruno Cardoso Lopes authored Aug 10, 2011
  278ffd7d
- Add 256-bit support for v8i32, v4i64 and v4f64 ISD::SELECT. Fix PR10556 · 72323966
  Bruno Cardoso Lopes authored Aug 09, 2011
  72323966
- Add v16i16 and v32i8 store patterns · fc481959
  Bruno Cardoso Lopes authored Aug 09, 2011
  fc481959
- Use fp unpack instructions to unpack int types. Until we have AVX2, this · 6963062a
  Bruno Cardoso Lopes authored Aug 09, 2011
  6963062a
- Fix a couple ridiculous copy-paste errors. rdar://9914773 . · 4ef2426b
  Eli Friedman authored Aug 09, 2011
  4ef2426b
Aug 09, 2011
- Reapply a more appropriate solution than in r137114. AVX supports · bed48dc8
  Bruno Cardoso Lopes authored Aug 09, 2011
  bed48dc8
- Revert r137114 · 24dd1d4a
  Bruno Cardoso Lopes authored Aug 09, 2011
  24dd1d4a
- Handle sitofp between v4f64 <- v4i32. Fix PR10559 · ad3453cf
  Bruno Cardoso Lopes authored Aug 09, 2011
  ad3453cf
- Add support for avx vector fextend · 1155b1ea
  Bruno Cardoso Lopes authored Aug 09, 2011
  1155b1ea
- Add AVX versions of 128-bit sitofp and fptosi · 0d0964d0
  Bruno Cardoso Lopes authored Aug 09, 2011
  0d0964d0
- Add two patterns to match special vmovss and vmovsd cases. Also fix · 2fc10736
  Bruno Cardoso Lopes authored Aug 09, 2011
  2fc10736
- Make LowerVSETCC aware of AVX types and add patterns to match them. · af6a8548
  Bruno Cardoso Lopes authored Aug 09, 2011
  af6a8548
Aug 08, 2011
- Add support for several vector shifts operations while in AVX mode. Fix PR10581 · c96953c1
  Bruno Cardoso Lopes authored Aug 08, 2011
  c96953c1
- Hoist hasLoadFromStackSlot and hasStoreToStackSlot. · daa2cad7
  Jakob Stoklund Olesen authored Aug 08, 2011
  daa2cad7