Commits · 97599f1211fea1fdc86784007762d2b8078ee7e3 · Roger Ferrer / llvm-epi-0.8

Mar 24, 2006
- plug the intrinsics into the patterns for movmsk* · 97599f12
  Chris Lattner authored Mar 24, 2006
```
llvm-svn: 27083
```
  97599f12
- Add dwarf register numbering to register data. · f0729b40
  Jim Laskey authored Mar 24, 2006
```
llvm-svn: 27081
```
  f0729b40
- Add support for dwarf register numbering. · 3b338d55
  Jim Laskey authored Mar 24, 2006
```
llvm-svn: 27080
```
  3b338d55
- add another note · 9f9b6116
  Chris Lattner authored Mar 24, 2006
```
llvm-svn: 27077
```
  9f9b6116
- add a note · 0affd761
  Chris Lattner authored Mar 24, 2006
```
llvm-svn: 27076
```
  0affd761
- Shuffle some includes around · c6b13e21
  Chris Lattner authored Mar 24, 2006
```
llvm-svn: 27073
```
  c6b13e21
- expose intrinsic info to the targets. · 58a96229
  Chris Lattner authored Mar 24, 2006
```
llvm-svn: 27070
```
  58a96229
- Fix a bad JIT encoding of VPERM. Why is VPERM D,A,B,C but vfmadd is D,A,C,B ?? · d589dd13
  Chris Lattner authored Mar 24, 2006
```
llvm-svn: 27069
```
  d589dd13
- Like the comment says, prefer to use the implicit add done by [r+r] addressing · f2286d59
  Chris Lattner authored Mar 24, 2006
```
modes than emitting an explicit add and using a base of r0.  This implements
Regression/CodeGen/PowerPC/mem-rr-addr-mode.ll

llvm-svn: 27068
```
  f2286d59
- Clean up some commentary. · 864e4447
  Jim Laskey authored Mar 24, 2006
```
llvm-svn: 27064
```
  864e4447
- Disable the i32->float G5 optimization. It is unsafe, as documented in the · a90b7141
  Chris Lattner authored Mar 24, 2006
```
comment.

This fixes 177.mesa, and McCat/09-vor with the td scheduler.

llvm-svn: 27060
```
  a90b7141
- add support for using vxor to build zero vectors. This implements · ab882abc
  Chris Lattner authored Mar 24, 2006
```
Regression/CodeGen/PowerPC/vec_zero.ll

llvm-svn: 27059
```
  ab882abc
- Handle BUILD_VECTOR with all zero elements. · 082c8785
  Evan Cheng authored Mar 24, 2006
```
llvm-svn: 27056
```
  082c8785
- Gabor points out that we can't spell. :) · f5efddf8
  Chris Lattner authored Mar 24, 2006
```
llvm-svn: 27049
```
  f5efddf8
- All v2f64 shuffle cases can be handled. · a91d8a5b
  Evan Cheng authored Mar 24, 2006
```
llvm-svn: 27044
```
  a91d8a5b
- More efficient v2f64 shuffle using movlhps, movhlps, unpckhpd, and unpcklpd. · 2595a687
  Evan Cheng authored Mar 24, 2006
```
llvm-svn: 27040
```
  2595a687
- A new entry · 6afb3c2d
  Evan Cheng authored Mar 24, 2006
```
llvm-svn: 27039
```
  6afb3c2d
- Ignore the burg output files. · f9c3dcfd
  Reid Spencer authored Mar 24, 2006
```
llvm-svn: 27033
```
  f9c3dcfd
- Handle more shuffle cases with SHUFP* instructions. · d27fb3e8
  Evan Cheng authored Mar 24, 2006
```
llvm-svn: 27024
```
  d27fb3e8
- Typo · 4b5b4e37
  Evan Cheng authored Mar 23, 2006
```
llvm-svn: 27008
```
  4b5b4e37
Mar 23, 2006
- add a note · cbcfe465
  Chris Lattner authored Mar 23, 2006
```
llvm-svn: 27000
```
  cbcfe465
- Typo · f842ea57
  Evan Cheng authored Mar 23, 2006
```
llvm-svn: 26997
```
  f842ea57
- Add PPC vector bit-convert support · 81137629
  Chris Lattner authored Mar 23, 2006
```
llvm-svn: 26995
```
  81137629
- Add support to locate local variables in frames (early version.) · 3c43609f
  Jim Laskey authored Mar 23, 2006
```
llvm-svn: 26994
```
  3c43609f
- Change interface to DwarfWriter. · cf0166fb
  Jim Laskey authored Mar 23, 2006
```
llvm-svn: 26991
```
  cf0166fb
- Modify how CBE handles #lines. · 267d39d1
  Jim Laskey authored Mar 23, 2006
```
llvm-svn: 26990
```
  267d39d1
- Fix the encodings of these new instructions, hopefully fixing the JIT · ce0206e1
  Chris Lattner authored Mar 23, 2006
```
failures from last night

llvm-svn: 26981
```
  ce0206e1
- Following icc's lead: use movdqa to load / store 128-bit integer vectors · 82ed4a42
  Evan Cheng authored Mar 23, 2006
```
llvm-svn: 26980
```
  82ed4a42
- Eliminate IntrinsicLowering from TargetMachine. · 6f95ab7a
  Chris Lattner authored Mar 23, 2006
```
Make the CBE and V9 backends create their own, since they're the only ones that use it.

llvm-svn: 26974
```
  6f95ab7a
- remove always-null IntrinsicLowering argument. · 811dd8d0
  Chris Lattner authored Mar 23, 2006
```
llvm-svn: 26971
```
  811dd8d0
- Add v4i32 <-> v4f32 bitconvert patterns. · 70558781
  Evan Cheng authored Mar 23, 2006
```
llvm-svn: 26969
```
  70558781
- Add 128-bit integer vector load and add (for testing). · b9b0550d
  Evan Cheng authored Mar 23, 2006
```
llvm-svn: 26967
```
  b9b0550d
- Add support for 8 bit immediates with 16/32 bit cmp instructions · fb6e0293
  Nate Begeman authored Mar 23, 2006
```
llvm-svn: 26966
```
  fb6e0293
Mar 22, 2006
- Added a ValueType operand to isShuffleMaskLegal(). For now, x86 will not do · 021bb7c9
  Evan Cheng authored Mar 22, 2006
```
64-bit vector shuffle.

llvm-svn: 26964
```
  021bb7c9
- SHUFP* are two address code. · ed794cd2
  Evan Cheng authored Mar 22, 2006
```
llvm-svn: 26959
```
  ed794cd2
- Some clean up. · bc047228
  Evan Cheng authored Mar 22, 2006
```
llvm-svn: 26957
```
  bc047228
- - Supposely movlhps is faster / better than unpcklpd. · d4e15579
  Evan Cheng authored Mar 22, 2006
```
- Don't forget pshufd is only available with sse2.

llvm-svn: 26956
```
  d4e15579
- - Implement X86ISelLowering::isShuffleMaskLegal(). We currently only support · 68ad48bd
  Evan Cheng authored Mar 22, 2006
```
  splat and PSHUFD cases.
- Clean up shuffle / splat matching code.

llvm-svn: 26954
```
  68ad48bd
- - VECTOR_SHUFFLE of v4i32 / v4f32 with undef second vector always matches · 8fdbdf20
  Evan Cheng authored Mar 22, 2006
```
  PSHUFD. We can make permutes entries which point to the undef pointing
  anything we want.
- Change some names to appease Chris.

llvm-svn: 26951
```
  8fdbdf20
- add a note · e24cf9df
  Chris Lattner authored Mar 22, 2006
```
llvm-svn: 26950
```
  e24cf9df