Added a movhlps, movlhps test case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27871 91177308-0d34-0410-b5e6-96231b3b80d8

Don't hardcode in 1.5 for the website, just use 'CVS'.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27870 91177308-0d34-0410-b5e6-96231b3b80d8

This is old, out of date, and isn't linked to by anything.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27869 91177308-0d34-0410-b5e6-96231b3b80d8

Make sure that the new instructions selected have the right type.  This fixes
CodeGen/PowerPC/2006-04-19-vmaddfp-crash.ll

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27868 91177308-0d34-0410-b5e6-96231b3b80d8

New testcase for a codegen crash

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27867 91177308-0d34-0410-b5e6-96231b3b80d8

Changing domain name

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27865 91177308-0d34-0410-b5e6-96231b3b80d8

Implement folding of a bunch of binops with undef

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27863 91177308-0d34-0410-b5e6-96231b3b80d8

Update llvmgcc4 tarball names

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27861 91177308-0d34-0410-b5e6-96231b3b80d8

Removed listing of llvm releases after 1.4, and said "1.4 and newer"

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27860 91177308-0d34-0410-b5e6-96231b3b80d8

Made warning red.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27859 91177308-0d34-0410-b5e6-96231b3b80d8

Document is out of date.. added warning and link to llvm-config.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27858 91177308-0d34-0410-b5e6-96231b3b80d8

Fixed up comment on xfail for llvmgcc version.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27857 91177308-0d34-0410-b5e6-96231b3b80d8

Added note about being able to XFAIL based on llvmgcc version.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27856 91177308-0d34-0410-b5e6-96231b3b80d8

Removed cvs mirror comment

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27855 91177308-0d34-0410-b5e6-96231b3b80d8

Minor fixes for the release.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27854 91177308-0d34-0410-b5e6-96231b3b80d8

This has been fixed!  Thanks Reid.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27853 91177308-0d34-0410-b5e6-96231b3b80d8

Yeah that's right!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27852 91177308-0d34-0410-b5e6-96231b3b80d8

Fixes from Tanya

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27851 91177308-0d34-0410-b5e6-96231b3b80d8

Add in missing #defines for _OpenBSD_ systems.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27850 91177308-0d34-0410-b5e6-96231b3b80d8

Handle v2i64 BUILD_VECTOR custom lowering correctly. v2i64 is a legal type,
but i64 is not. If possible, change a i64 op to a f64 (e.g. load, constant)
and then cast it back.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27849 91177308-0d34-0410-b5e6-96231b3b80d8

Allow OpenBSD to be recognized as a UNIX platform.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27848 91177308-0d34-0410-b5e6-96231b3b80d8

isSplatMask() bug: first element can be an undef.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27847 91177308-0d34-0410-b5e6-96231b3b80d8

Simplify some code

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27846 91177308-0d34-0410-b5e6-96231b3b80d8

- Added support to do aribitrary 4 wide shuffle with no more than three
  instructions.
- Fixed a commute vector_shuff bug.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27845 91177308-0d34-0410-b5e6-96231b3b80d8

Prefer {p}unpack* and mov*dup over {p}shuf* as well.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27844 91177308-0d34-0410-b5e6-96231b3b80d8

Renamed AddedCost to AddedComplexity.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27843 91177308-0d34-0410-b5e6-96231b3b80d8

- Renamed AddedCost to AddedComplexity.
- Added more movhlps and movlhps patterns.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27842 91177308-0d34-0410-b5e6-96231b3b80d8

Rename AddedCost to AddedComplexity.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27841 91177308-0d34-0410-b5e6-96231b3b80d8

Commute vector_shuffle to match more movlhps, movlp{s|d} cases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27840 91177308-0d34-0410-b5e6-96231b3b80d8

Final piece to get relinked .o files buildable universal on Darwin.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27839 91177308-0d34-0410-b5e6-96231b3b80d8

Regenerate

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27838 91177308-0d34-0410-b5e6-96231b3b80d8

When on darwin, compiler_flags need to be percolated down to the 'gcc -r'
command line so that relinked .o files can be built universal.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27837 91177308-0d34-0410-b5e6-96231b3b80d8

More mov{h|l}p{d|s} patterns.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27836 91177308-0d34-0410-b5e6-96231b3b80d8

- More mov{h|l}ps patterns.
- Increase cost (complexity) of patterns which match mov{h|l}ps ops. These
  are preferred over shufps in most cases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27835 91177308-0d34-0410-b5e6-96231b3b80d8

Allow "let AddedCost = n in" to increase pattern complexity.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27834 91177308-0d34-0410-b5e6-96231b3b80d8

Alpha too!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27833 91177308-0d34-0410-b5e6-96231b3b80d8

add a note

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27832 91177308-0d34-0410-b5e6-96231b3b80d8

Another simple case type merge case to try

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27831 91177308-0d34-0410-b5e6-96231b3b80d8

deal with memchr

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27830 91177308-0d34-0410-b5e6-96231b3b80d8

friendlier error message

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27829 91177308-0d34-0410-b5e6-96231b3b80d8

add a note

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27828 91177308-0d34-0410-b5e6-96231b3b80d8

Add a note.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27827 91177308-0d34-0410-b5e6-96231b3b80d8

grammaro

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27826 91177308-0d34-0410-b5e6-96231b3b80d8

Fix a bug owen noticed

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27825 91177308-0d34-0410-b5e6-96231b3b80d8

Change wording

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27824 91177308-0d34-0410-b5e6-96231b3b80d8

add a note

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27823 91177308-0d34-0410-b5e6-96231b3b80d8

add some more notes

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27822 91177308-0d34-0410-b5e6-96231b3b80d8

stupid stuff

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27821 91177308-0d34-0410-b5e6-96231b3b80d8

fix printing call graphs

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27820 91177308-0d34-0410-b5e6-96231b3b80d8

I understand now.  Shoot.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27819 91177308-0d34-0410-b5e6-96231b3b80d8

- PEXTRW cannot take a memory location as its first source operand.
- PINSRWrmi encoding bug.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27818 91177308-0d34-0410-b5e6-96231b3b80d8

SHUFP{S|D}, PSHUF* encoding bugs. Left out the mask immediate operand.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27817 91177308-0d34-0410-b5e6-96231b3b80d8

Name change for clarity sake

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27816 91177308-0d34-0410-b5e6-96231b3b80d8

Encoding bug: CMPPSrmi, CMPPDrmi dropped operand 2 (condtion immediate).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27815 91177308-0d34-0410-b5e6-96231b3b80d8

Name change for clarity sake

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27814 91177308-0d34-0410-b5e6-96231b3b80d8

Left a pattern out

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27813 91177308-0d34-0410-b5e6-96231b3b80d8

llvm.memc* improvements.  helps PA a lot in some specmarks

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27812 91177308-0d34-0410-b5e6-96231b3b80d8

llvm.memc* improvements.  helps PA a lot in some specmarks

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27811 91177308-0d34-0410-b5e6-96231b3b80d8

These are correctly encoded by the JIT.  I checked :)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27810 91177308-0d34-0410-b5e6-96231b3b80d8

add a note

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27809 91177308-0d34-0410-b5e6-96231b3b80d8

Fix a crash on:
void foo2(vector float *A, vector float *B) {
  vector float C = (vector float)vec_cmpeq(*A, *B);
  if (!vec_any_eq(*A, *B))
    *B = (vector float){0,0,0,0};
  *A = C;
}

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27808 91177308-0d34-0410-b5e6-96231b3b80d8

Fixed an encoding bug: movd from XMM to R32.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27807 91177308-0d34-0410-b5e6-96231b3b80d8

pretty print node name

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27806 91177308-0d34-0410-b5e6-96231b3b80d8

Implement an important entry from README_ALTIVEC:

If an altivec predicate compare is used immediately by a branch, don't
use a (serializing) MFCR instruction to read the CR6 register, which requires
a compare to get it back to CR's.  Instead, just branch on CR6 directly. :)

For example, for:
void foo2(vector float *A, vector float *B) {
  if (!vec_any_eq(*A, *B))
    *B = (vector float){0,0,0,0};
}

We now generate:

_foo2:
        mfspr r2, 256
        oris r5, r2, 12288
        mtspr 256, r5
        lvx v2, 0, r4
        lvx v3, 0, r3
        vcmpeqfp. v2, v3, v2
        bne cr6, LBB1_2 ; UnifiedReturnBlock
LBB1_1: ; cond_true
        vxor v2, v2, v2
        stvx v2, 0, r4
        mtspr 256, r2
        blr
LBB1_2: ; UnifiedReturnBlock
        mtspr 256, r2
        blr

instead of:

_foo2:
        mfspr r2, 256
        oris r5, r2, 12288
        mtspr 256, r5
        lvx v2, 0, r4
        lvx v3, 0, r3
        vcmpeqfp. v2, v3, v2
        mfcr r3, 2
        rlwinm r3, r3, 27, 31, 31
        cmpwi cr0, r3, 0
        beq cr0, LBB1_2 ; UnifiedReturnBlock
LBB1_1: ; cond_true
        vxor v2, v2, v2
        stvx v2, 0, r4
        mtspr 256, r2
        blr
LBB1_2: ; UnifiedReturnBlock
        mtspr 256, r2
        blr

This implements CodeGen/PowerPC/vec_br_cmp.ll.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27804 91177308-0d34-0410-b5e6-96231b3b80d8

new testcase

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27803 91177308-0d34-0410-b5e6-96231b3b80d8

move some stuff around, clean things up

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27802 91177308-0d34-0410-b5e6-96231b3b80d8

Teach the codegen about instructions used for SSE spill code, allowing it
to optimize cases where it has to spill a lot

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27801 91177308-0d34-0410-b5e6-96231b3b80d8

Fix a copy & paste error from long ago.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27800 91177308-0d34-0410-b5e6-96231b3b80d8

Add some more notes, many still missing

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27799 91177308-0d34-0410-b5e6-96231b3b80d8

Have the AutoRegen.sh script prompt the user for the LLVM src and obj
directories if it can't find them. Then, replace those values into the
configure.ac script and pass them to the LLVM_CONFIG_PROJECT so that the
values become the default for llvm_src and llvm_obj variables. In this way
the user is required to input this exactly once, and the scripts take it
from there.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27798 91177308-0d34-0410-b5e6-96231b3b80d8

Make it possible to default the llvm_src and llvm_obj variables based on
the arguments to the macro. This better supports the AutoRegen.sh script
in projects/sample/autoconf.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27797 91177308-0d34-0410-b5e6-96231b3b80d8

add a bunch of stuff, pieces still missing

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27796 91177308-0d34-0410-b5e6-96231b3b80d8

Add a warning.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27795 91177308-0d34-0410-b5e6-96231b3b80d8

Add a warning

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27794 91177308-0d34-0410-b5e6-96231b3b80d8

Use vmladduhm to do v8i16 multiplies which is faster and simpler than doing
even/odd halves.  Thanks to Nate telling me what's what.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27793 91177308-0d34-0410-b5e6-96231b3b80d8

Implement v16i8 multiply with this code:

        vmuloub v5, v3, v2
        vmuleub v2, v3, v2
        vperm v2, v2, v5, v4

This implements CodeGen/PowerPC/vec_mul.ll.  With this, v16i8 multiplies are
6.79x faster than before.

Overall, UnitTests/Vector/multiplies.c is now 2.45x faster with LLVM than with
GCC.

Remove the 'integer multiplies' todo from the README file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27792 91177308-0d34-0410-b5e6-96231b3b80d8

Add tests for v8i16 and v16i8

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27791 91177308-0d34-0410-b5e6-96231b3b80d8

Correct comments

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27790 91177308-0d34-0410-b5e6-96231b3b80d8

Lower v8i16 multiply into this code:

        li r5, lo16(LCPI1_0)
        lis r6, ha16(LCPI1_0)
        lvx v4, r6, r5
        vmulouh v5, v3, v2
        vmuleuh v2, v3, v2
        vperm v2, v2, v5, v4

where v4 is:
LCPI1_0:                                        ;  <16 x ubyte>
        .byte   2
        .byte   3
        .byte   18
        .byte   19
        .byte   6
        .byte   7
        .byte   22
        .byte   23
        .byte   10
        .byte   11
        .byte   26
        .byte   27
        .byte   14
        .byte   15
        .byte   30
        .byte   31

This is 5.07x faster on the G5 (measured) than lowering to scalar code +
loads/stores.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27789 91177308-0d34-0410-b5e6-96231b3b80d8

Custom lower v4i32 multiplies into a cute sequence, instead of having legalize
scalarize the sequence into 4 mullw's and a bunch of load/store traffic.

This speeds up v4i32 multiplies 4.1x (measured) on a G5.  This implements
PowerPC/vec_mul.ll

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27788 91177308-0d34-0410-b5e6-96231b3b80d8

new testcase

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27787 91177308-0d34-0410-b5e6-96231b3b80d8

Another entry

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27786 91177308-0d34-0410-b5e6-96231b3b80d8

Fix a build failure on Vladimir's tester.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27785 91177308-0d34-0410-b5e6-96231b3b80d8

Another entry.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27784 91177308-0d34-0410-b5e6-96231b3b80d8

Use movss to insert_vector_elt(v, s, 0).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27782 91177308-0d34-0410-b5e6-96231b3b80d8

Turn x86 unaligned load/store intrinsics into aligned load/store instructions
if the pointer is known aligned.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27781 91177308-0d34-0410-b5e6-96231b3b80d8

Fix handling of calls in functions that use vectors.  This fixes a crash on
the code in GCC PR26546.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27780 91177308-0d34-0410-b5e6-96231b3b80d8

Use two pinsrw to insert an element into v4i32 / v4f32 vector.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27779 91177308-0d34-0410-b5e6-96231b3b80d8

remove done item

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27778 91177308-0d34-0410-b5e6-96231b3b80d8

Don't diddle VRSAVE if no registers need to be added/removed from it.  This
allows us to codegen functions as:

_test_rol:
        vspltisw v2, -12
        vrlw v2, v2, v2
        blr

instead of:

_test_rol:
        mfvrsave r2, 256
        mr r3, r2
        mtvrsave r3
        vspltisw v2, -12
        vrlw v2, v2, v2
        mtvrsave r2
        blr

Testcase here: CodeGen/PowerPC/vec_vrsave.ll

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27777 91177308-0d34-0410-b5e6-96231b3b80d8

New testcase, shouldn't touch vrsave

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27776 91177308-0d34-0410-b5e6-96231b3b80d8

Add a MachineInstr::eraseFromParent convenience method.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27775 91177308-0d34-0410-b5e6-96231b3b80d8

Add some convenience methods.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27774 91177308-0d34-0410-b5e6-96231b3b80d8

Encoding bug

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27773 91177308-0d34-0410-b5e6-96231b3b80d8

Vectors that are known live-in and live-out are clearly already marked in
the vrsave register for the caller.  This allows us to codegen a function as:

_test_rol:
        mfspr r2, 256
        mr r3, r2
        mtspr 256, r3
        vspltisw v2, -12
        vrlw v2, v2, v2
        mtspr 256, r2
        blr

instead of:

_test_rol:
        mfspr r2, 256
        oris r3, r2, 40960
        mtspr 256, r3
        vspltisw v0, -12
        vrlw v2, v0, v0
        mtspr 256, r2
        blr

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27772 91177308-0d34-0410-b5e6-96231b3b80d8

Prefer to allocate V2-V5 before V0,V1.  This lets us generate code like this:

        vspltisw v2, -12
        vrlw v2, v2, v2

instead of:

        vspltisw v0, -12
        vrlw v2, v0, v0

when a function is returning a value.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27771 91177308-0d34-0410-b5e6-96231b3b80d8

Move some knowledge about registers out of the code emitter into the register info.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27770 91177308-0d34-0410-b5e6-96231b3b80d8

Use a small table instead of macros to do this conversion.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27769 91177308-0d34-0410-b5e6-96231b3b80d8

Implement v8i16, v16i8 splat using unpckl + pshufd.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27768 91177308-0d34-0410-b5e6-96231b3b80d8

implement returns of a vector, testcase here: CodeGen/X86/vec_return.ll

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27767 91177308-0d34-0410-b5e6-96231b3b80d8