6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-08-02 09:29:24 +00:00
Code Issues Projects Releases Wiki Activity
24,371 Commits 39 Branches 0 Tags 277 MiB
a083af14c8
Commit Graph

181 Commits

Author SHA1 Message Date
Evan Cheng
a083af14c8 Fix a performance regression. Use {p}shuf* when there are only two distinct elements in a build_vector. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27945 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-22 06:21:46 +00:00
Evan Cheng
ba05f728b5 Revamp build_vector lowering to take advantage of movss and movd instructions. 
movd always clear the top 96 bits and movss does so when it's loading the
value from memory.
The net result is codegen for 4-wide shuffles is much improved. It is near
optimal if one or more elements is a zero. e.g.

__m128i test(int a, int b) {
  return _mm_set_epi32(0, 0, b, a);
}

compiles to

_test:
	movd 8(%esp), %xmm1
	movd 4(%esp), %xmm0
	punpckldq %xmm1, %xmm0
	ret

compare to gcc:

_test:
	subl	$12, %esp
	movd	20(%esp), %xmm0
	movd	16(%esp), %xmm1
	punpckldq	%xmm0, %xmm1
	movq	%xmm1, %xmm0
	movhps	LC0, %xmm0
	addl	$12, %esp
	ret

or icc:

_test:
        movd      4(%esp), %xmm0                                #5.10
        movd      8(%esp), %xmm3                                #5.10
        xorl      %eax, %eax                                    #5.10
        movd      %eax, %xmm1                                   #5.10
        punpckldq %xmm1, %xmm0                                  #5.10
        movd      %eax, %xmm2                                   #5.10
        punpckldq %xmm2, %xmm3                                  #5.10
        punpckldq %xmm3, %xmm0                                  #5.10
        ret                                                     #5.10

There are still room for improvement, for example the FP variant of the above example:

__m128 test(float a, float b) {
  return _mm_set_ps(0.0, 0.0, b, a);
}

_test:
	movss 8(%esp), %xmm1
	movss 4(%esp), %xmm0
	unpcklps %xmm1, %xmm0
	xorps %xmm1, %xmm1
	movlhps %xmm1, %xmm0
	ret

The xorps and movlhps are unnecessary. This will require post legalizer optimization to handle.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27939 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-21 23:03:30 +00:00
Evan Cheng
017dcc6e55 Now generating perfect (I think) code for "vector set" with a single non-zero 
scalar value.

e.g.
        _mm_set_epi32(0, a, 0, 0);
==>
	movd 4(%esp), %xmm0
	pshufd $69, %xmm0, %xmm0

        _mm_set_epi8(0, 0, 0, 0, 0, a, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
==>
	movzbw 4(%esp), %ax
	movzwl %ax, %eax
	pxor %xmm0, %xmm0
	pinsrw $5, %eax, %xmm0


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27923 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-21 01:05:10 +00:00
Evan Cheng
39623daef6 - Added support to turn "vector clear elements", e.g. pand V, <-1, -1, 0, -1> 
to a vector shuffle.
- VECTOR_SHUFFLE lowering change in preparation for more efficient codegen
of vector shuffle with zero (or any splat) vector.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27875 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-20 08:58:49 +00:00
Evan Cheng
72cd9a9439 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
 2006-04-20 00:11:39 +00:00
Evan Cheng
94fe5eb14a 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
 2006-04-19 23:28:59 +00:00
Evan Cheng
80d428c370 - 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
 2006-04-19 22:48:17 +00:00
Evan Cheng
533a0aa9ba 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
 2006-04-19 20:35:22 +00:00
Evan Cheng
cdfc3c82a7 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
 2006-04-17 22:45:49 +00:00
Evan Cheng
5edb8d270c 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
 2006-04-17 22:04:06 +00:00
Evan Cheng
c575ca22ea Implement v8i16, v16i8 splat using unpckl + pshufd. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27768 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-17 20:43:08 +00:00
Chris Lattner
b2be4032c5 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
 2006-04-17 20:32:50 +00:00
Evan Cheng
5001ea1078 FP SETOLT, SETOLT, SETUGE, SETUGT conditions were implemented incorrectly 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27755 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-17 07:24:10 +00:00
Evan Cheng
57ebe9fbf0 Silly bug 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27719 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-15 05:37:34 +00:00
Evan Cheng
39fc145995 Do not use movs{h|l}dup for a shuffle with a single non-undef node. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27718 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-15 03:13:24 +00:00
Evan Cheng
d953947d26 Last few SSE3 intrinsics. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27711 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-14 21:59:03 +00:00
Evan Cheng
f99898453d X86 SSE2 supports v8i16 multiplication 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27644 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-13 05:10:25 +00:00
Evan Cheng
2c3ae37213 All "integer" logical ops (pand, por, pxor) are now promoted to v2i64. 
Clean up and fix various logical ops issues.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27633 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-12 21:21:57 +00:00
Evan Cheng
91b740da12 Promote v4i32, v8i16, v16i8 load to v2i64 load. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27612 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-12 17:12:36 +00:00
Evan Cheng
d6d1cbd692 Added support for _mm_move_ss and _mm_move_sd. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27575 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-11 00:19:04 +00:00
Evan Cheng
f7c378e9ea Conditional move of vector types. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27556 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-10 07:23:14 +00:00
Evan Cheng
c5cdff2341 Code clean up. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27501 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-07 21:53:05 +00:00
Evan Cheng
5ced1d812e - movlp{s|d} and movhp{s|d} support. 
- Normalize shuffle nodes so result vector lower half elements come from the
  first vector, the rest come from the second vector. (Except for the
  exceptions :-).
- Other minor fixes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27474 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-06 23:23:56 +00:00
Evan Cheng
6be2c58c8c Support for comi / ucomi intrinsics. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27444 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-05 23:38:46 +00:00
Evan Cheng
1d5a8cca00 Handle canonical form of e.g. 
vector_shuffle v1, v1, <0, 4, 1, 5, 2, 6, 3, 7>

This is turned into
vector_shuffle v1, <undef>, <0, 0, 1, 1, 2, 2, 3, 3>
by dag combiner.

It would match a {p}unpckl on x86.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27437 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-05 07:20:06 +00:00
Evan Cheng
865f0606f7 Bogus assert 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27434 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-05 06:11:20 +00:00
Evan Cheng
278158b487 Fallthrough to expand if a VECTOR_SHUFFLE cannot be custom lowered. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27433 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-05 06:09:26 +00:00
Evan Cheng
c21a053729 Handle v8i16 shuffle that must be broken into a pair of pshufhw / pshuflw. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27427 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-05 01:47:37 +00:00
Evan Cheng
20e3ed102b Use movlpd to: store lower f64 extracted from v2f64. 
Use movhpd to: store upper f64 extracted from v2f64.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27382 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-03 22:30:54 +00:00
Evan Cheng
11e15b38e9 - More efficient extract_vector_elt with shuffle and movss, movsd, movd, etc. 
- Some bug fixes and naming inconsistency fixes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27377 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-04-03 20:53:28 +00:00
Evan Cheng
653159f4aa Use a X86 target specific node X86ISD::PINSRW instead of a mal-formed 
INSERT_VECTOR_ELT to insert a 16-bit value in a 128-bit vector.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27314 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-31 21:55:24 +00:00
Evan Cheng
b067a1e7e6 Add support to use pextrw and pinsrw to extract and insert a word element 
from a 128-bit vector.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27304 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-31 19:22:53 +00:00
Evan Cheng
33e85ca7b6 Expand all INSERT_VECTOR_ELT (obviously bad) for now. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27275 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-31 01:30:39 +00:00
Evan Cheng
fb47a9b1c8 Typo 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27272 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-31 00:33:57 +00:00
Evan Cheng
ef698ca30d Ok for vector_shuffle mask to contain undef elements. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27271 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-31 00:30:29 +00:00
Evan Cheng
7d9061e300 Make sure all possible shuffles are matched. 
Use pshufd, pshuhw, and pshulw to shuffle v4f32 if shufps doesn't match.
Use shufps to shuffle v4f32 if pshufd, pshuhw, and pshulw don't match.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27259 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-30 19:54:57 +00:00
Evan Cheng
506d3dfa90 - Added some SSE2 128-bit packed integer ops. 
- Added SSE2 128-bit integer pack with signed saturation ops.
- Added pshufhw and pshuflw ops.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27252 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-29 23:07:14 +00:00
Evan Cheng
691c923e47 Need to special case splat after all. Make the second operand of splat 
vector_shuffle undef.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27250 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-29 19:02:40 +00:00
Evan Cheng
475aecf467 - More shuffle related bug fixes. 
- Whenever possible use ops of the right packed types for vector shuffles /
  splats.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27246 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-29 03:04:49 +00:00
Evan Cheng
4f5633883b - Only use pshufd for v4i32 vector shuffles. 
- Other shuffle related fixes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27244 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-29 01:30:51 +00:00
Evan Cheng
36b27f3cde Fixing buggy code. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27239 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-28 23:41:33 +00:00
Jim Laskey
2d2a6131da Added missing paren on behalf of Ramana Radhakrishnan. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27223 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-28 10:17:11 +00:00
Evan Cheng
ed4ca7f6c3 Missed X86::isUNPCKHMask 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27222 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-28 08:27:15 +00:00
Evan Cheng
2064a2b47e * Prefer using operation of matching types. e.g unpcklpd rather than movlhps. 
* Bug fixes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27218 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-28 06:50:32 +00:00
Evan Cheng
4fcb922c70 - Clean up / consoladate various shuffle masks. 
- Some misc. bug fixes.
- Use MOVHPDrm to load from m64 to upper half of a XMM register.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27210 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-28 02:43:26 +00:00
Evan Cheng
0038e59803 Model unpack lower and interleave as vector_shuffle so we can lower the 
intrinsics as such.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27200 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-28 00:39:58 +00:00
Evan Cheng
a0b3afbe14 Use pcmpeq to generate vector of all ones. 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27167 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-27 07:00:16 +00:00
Nate Begeman
f15485a8d0 SelectionDAGISel can now natively handle Switch instructions, in the same 
manner that the LowerSwitch LLVM to LLVM pass does: emitting a binary
search tree of basic blocks.  The new approach has several advantages:
it is faster, it generates significantly smaller code in many cases, and
it paves the way for implementing dense switch tables as a jump table by
handling switches directly in the instruction selector.

This functionality is currently only enabled on x86, but should be safe for
every target.  In anticipation of making it the default, the cfg is now
properly updated in the x86, ppc, and sparc select lowering code.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27156 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-27 01:32:24 +00:00
Evan Cheng
ffea91e522 Remove X86:isZeroVector, use ISD::isBuildVectorAllZeros instead; some fixes / cleanups 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27150 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-26 09:53:12 +00:00
Evan Cheng
c60bd97b94 Build arbitrary vector with more than 2 distinct scalar elements with a 
series of unpack and interleave ops.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27119 91177308-0d34-0410-b5e6-96231b3b80d8
 2006-03-25 09:37:23 +00:00