Commits · 2771e2c9602df0f3e296e18500b9079b21dea98c · Roger Ferrer / llvm-epi-0.8

Mar 25, 2006

Chris Lattner authored Mar 25, 2006

 <int -1, int -1, int -1, int -1>
and
 <int 65537, int 65537, int 65537, int 65537>

Using things like:
  vspltisb v0, -1
and:
  vspltish v0, 1

instead of using constant pool loads.

This implements CodeGen/PowerPC/vec_splat.ll:splat_imm_i{32|16}.

llvm-svn: 27106

2771e2c9

Mar 24, 2006
- Add dwarf register numbering to register data. · f0729b40
  Jim Laskey authored Mar 24, 2006
```
llvm-svn: 27081
```
  f0729b40
- add another note · 9f9b6116
  Chris Lattner authored Mar 24, 2006
```
llvm-svn: 27077
```
  9f9b6116
- 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
- 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
- Gabor points out that we can't spell. :) · f5efddf8
  Chris Lattner authored Mar 24, 2006
```
llvm-svn: 27049
```
  f5efddf8
Mar 23, 2006
- add a note · cbcfe465
  Chris Lattner authored Mar 23, 2006
```
llvm-svn: 27000
```
  cbcfe465
- 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
- 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
Mar 22, 2006

This has been implemented. Tweak it into another note · eccf4695
Chris Lattner authored Mar 22, 2006
```
llvm-svn: 26944
```
eccf4695

When possible, custom lower 32-bit SINT_TO_FP to this: · 4a66d694

Chris Lattner authored Mar 22, 2006

_foo2:
        extsw r2, r3
        std r2, -8(r1)
        lfd f0, -8(r1)
        fcfid f0, f0
        frsp f1, f0
        blr

instead of this:

_foo2:
        lis r2, ha16(LCPI2_0)
        lis r4, 17200
        xoris r3, r3, 32768
        stw r3, -4(r1)
        stw r4, -8(r1)
        lfs f0, lo16(LCPI2_0)(r2)
        lfd f1, -8(r1)
        fsub f0, f1, f0
        frsp f1, f0
        blr

This speeds up Misc/pi from 2.44s->2.09s with LLC and from 3.01->2.18s
with llcbeta (16.7% and 38.1% respectively).

llvm-svn: 26943

4a66d694

Add support for "ri" addressing modes where the immediate is a 14-bit field · 77373d1b
Chris Lattner authored Mar 22, 2006
```
which is shifted left two bits before use.  Instructions like STD use this
addressing mode.

llvm-svn: 26942
```
77373d1b
Fix the JIT encoding of the VAForm_1 instructions, including vmaddfp · 4e737175
Chris Lattner authored Mar 22, 2006
```
llvm-svn: 26935
```
4e737175

Mar 21, 2006
- These targets don't support EXTRACT_VECTOR_ELT, though, in time, X86 will. · 00f4683b
  Chris Lattner authored Mar 21, 2006
```
llvm-svn: 26930
```
  00f4683b
- Don't emit pseudo instructions! · 3a2ae6ad
  Chris Lattner authored Mar 21, 2006
```
llvm-svn: 26926
```
  3a2ae6ad
- Update readme · 01312798
  Nate Begeman authored Mar 21, 2006
```
llvm-svn: 26924
```
  01312798
- Print absolute memory references like this: · 139eac5b
  Chris Lattner authored Mar 21, 2006
```
       lwz r2, 8(0)
instead of this:
       lwz r2, 8(r0)

This fixes the llc/llc-beta failures on PPC last night.

llvm-svn: 26922
```
  139eac5b
- With Evan's latest tblgen patch, this code is obsolete, thanks Evan! · bda7310e
  Chris Lattner authored Mar 21, 2006
```
llvm-svn: 26917
```
  bda7310e
- When codegen'ing vector MUL using VFMADD, *add* the 0, don't *mul* the 0. · d2132f87
  Chris Lattner authored Mar 21, 2006
```
llvm-svn: 26913
```
  d2132f87
- minor note · f1948341
  Chris Lattner authored Mar 21, 2006
```
llvm-svn: 26912
```
  f1948341
Mar 20, 2006
- Handle constant addresses more efficiently, folding the low bits into the · c8b16d00
  Chris Lattner authored Mar 20, 2006
```
disp field of the load/store if possible.  This compiles
CodeGen/PowerPC/load-constant-addr.ll to:

_test:
        lis r2, 2838
        lfs f1, 26848(r2)
        blr

instead of:

_test:
        lis r2, 2838
        ori r2, r2, 26848
        lfs f1, 0(r2)
        blr

llvm-svn: 26908
```
  c8b16d00
- remove dead variable · 6d74b09d
  Chris Lattner authored Mar 20, 2006
```
llvm-svn: 26907
```
  6d74b09d
- Fix a couple of bugs in permute/splat generate, thanks to Nate for actually · a1bc294f
  Chris Lattner authored Mar 20, 2006
```
figuring these out! :)

llvm-svn: 26904
```
  a1bc294f
- reenable this hack, the tblgen version isn't quite ready · eda030da
  Chris Lattner authored Mar 20, 2006
```
llvm-svn: 26902
```
  eda030da
- Fix the pattern for VADDUWM, add i32 splat · f96d523b
  Chris Lattner authored Mar 20, 2006
```
llvm-svn: 26901
```
  f96d523b
- Use tblgen'd VECTOR_SHUFFLE selection code. · 89f3cff0
  Evan Cheng authored Mar 20, 2006
```
llvm-svn: 26900
```
  89f3cff0
- Add support for generating vspltw, instead of a vperm instruction with a · a9a13133
  Chris Lattner authored Mar 20, 2006
```
constant pool load.  This generates significantly nicer code for splats.

When tblgen gets bugfixed, we can remove the custom selection code.

llvm-svn: 26898
```
  a9a13133
- Implement PPC::isSplatShuffleMask and PPC::getVSPLTImmediate. · a8fbb6dd
  Chris Lattner authored Mar 20, 2006
```
llvm-svn: 26897
```
  a8fbb6dd
- fix duplicate definition errors · ffc47568
  Chris Lattner authored Mar 20, 2006
```
llvm-svn: 26896
```
  ffc47568
- Check in some intermediate code that adds a skeleton for matching vsplt* · 382f356b
  Chris Lattner authored Mar 20, 2006
```
instructions

llvm-svn: 26894
```
  382f356b
- fix typo · 93d99f99
  Chris Lattner authored Mar 20, 2006
```
llvm-svn: 26889
```
  93d99f99
- add vsplat instructions, fix sched description for vperm · 366b2514
  Chris Lattner authored Mar 20, 2006
```
llvm-svn: 26888
```
  366b2514
- Custom lower arbitrary VECTOR_SHUFFLE's to VPERM. · a8713b1e
  Chris Lattner authored Mar 20, 2006
```
TODO: leave specific ones as VECTOR_SHUFFLE's and turn them into specialized
operations like vsplt*

llvm-svn: 26887
```
  a8713b1e
- Claim to have v16i8 for perm masks · 0a8b4eae
  Chris Lattner authored Mar 20, 2006
```
llvm-svn: 26886
```
  0a8b4eae
- add the vperm instruction · e7a058de
  Chris Lattner authored Mar 20, 2006
```
llvm-svn: 26883
```
  e7a058de
Mar 19, 2006
- Add a note about the MUL -> FMADD vector bug. · 169e6238
  Chris Lattner authored Mar 19, 2006
```
llvm-svn: 26874
```
  169e6238