maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
65,009 Commits 23 Branches 0 Tags 2.2 GiB
Commit Graph

331 Commits

Author SHA1 Message Date
Dan Gohman 29845cd40d Fix the spelling of the prefetchnta instruction. 
llvm-svn: 36256
 2007-04-18 14:09:14 +00:00
Bill Wendling f099841573 Add support for our first SSSE3 instruction "pmulhrsw". 
llvm-svn: 35869
 2007-04-10 22:10:25 +00:00
Evan Cheng 61eee86487 Mark re-materializable instructions. 
llvm-svn: 35230
 2007-03-21 00:16:56 +00:00
Chris Lattner d647f92a14 add missing braces 
llvm-svn: 34905
 2007-03-04 06:13:52 +00:00
Evan Cheng b68a774fd9 How the heck did I forget patterns for llvm.x86.sse2.cmp.sd? 
llvm-svn: 34434
 2007-02-20 00:39:09 +00:00
Evan Cheng 82241c86e9 - FCOPYSIGN custom lowering bug. Clear the sign bit of operand 0 first before 
or'ing in the sign bit of operand 1.
- Tweaking: rather than left shift the sign bit, fp_extend operand 1 first
  before taking its sign bit if its type is smaller than that of operand 0.

llvm-svn: 32932
 2007-01-05 21:37:56 +00:00
Evan Cheng 4363e884c0 With SSE2, expand FCOPYSIGN to a series of SSE bitwise operations. 
llvm-svn: 32900
 2007-01-05 07:55:56 +00:00
Evan Cheng fccea9b2a9 - Rename MOVDSS2DIrr to MOVSS2DIrr for consistency sake. 
- Add MOVDI2SSrm and MOVSS2DImr to fold load / store for i32 <-> f32 bit_convert
  patterns.

llvm-svn: 32582
 2006-12-14 19:43:11 +00:00
Chris Lattner c20b7e878a If we have ScalarSSE, we can select bitconvert into single instructions. 
This compiles bitcast.ll:test3/test4 into:

_test3:
        movd %xmm0, %eax
        ret
_test4:
        movd %edi, %xmm0
        ret

llvm-svn: 32230
 2006-12-05 18:45:06 +00:00
Evan Cheng 572dc9cb4e Correct instructions for moving data between GR64 and SSE registers; also correct load i64 / store i64 from v2i64. 
llvm-svn: 31795
 2006-11-16 23:33:25 +00:00
Evan Cheng 49683ba236 Don't dag combine floating point select to max and min intrinsics. Those 
take v4f32 / v2f64 operands and may end up causing larger spills / restores.
Added X86 specific nodes X86ISD::FMAX, X86ISD::FMIN instead.

This fixes PR996.

llvm-svn: 31645
 2006-11-10 21:43:37 +00:00
Evan Cheng 922e191116 Fixed a bug which causes x86 be to incorrectly match 
shuffle v, undef, <2, ?, 3, ?>
to movhlps
It should match to unpckhps instead.

Added proper matching code for
shuffle v, undef, <2, 3, 2, 3>

llvm-svn: 31519
 2006-11-07 22:14:24 +00:00
Chris Lattner 9ac6442db6 remove dead/redundant vars 
llvm-svn: 31435
 2006-11-03 23:48:56 +00:00
Evan Cheng 94e5bc9e83 Fix ldmxcsr JIT encoding. 
llvm-svn: 31343
 2006-11-01 06:53:52 +00:00
Evan Cheng e056dd5928 Fixed a significant bug where unpcklpd is incorrectly used to extract element 1 from a v2f64 value. 
llvm-svn: 31228
 2006-10-27 21:08:32 +00:00
Evan Cheng 7e065ff7e4 X86ISD::PEXTRW 3rd operand type is always target pointer type. 
llvm-svn: 31185
 2006-10-25 21:35:05 +00:00
Evan Cheng 4090dc4703 ComplexPatterns sse_load_f32 and sse_load_f64 returns in / out chain operands. 
llvm-svn: 30892
 2006-10-11 21:06:01 +00:00
Evan Cheng 57ccb6d372 Don't go too crazy with these AddComplexity. Try matching shufps with load 
folding first.

llvm-svn: 30848
 2006-10-09 21:42:15 +00:00
Evan Cheng e71fe34d75 Reflects ISD::LOAD / ISD::LOADX / LoadSDNode changes. 
llvm-svn: 30844
 2006-10-09 20:57:25 +00:00
Chris Lattner 398195ebbe completely disable folding of loads into scalar sse instructions and provide 
a framework for doing it right.  This fixes
CodeGen/X86/2006-10-07-ScalarSSEMiscompile.ll.

Once X86DAGToDAGISel::SelectScalarSSELoad is implemented right, this task
will be done.

llvm-svn: 30817
 2006-10-07 21:55:32 +00:00
Chris Lattner 942009fee5 convert packed FP add/sub/mul/div to use a multiclass. 
llvm-svn: 30815
 2006-10-07 21:17:13 +00:00
Chris Lattner 4005f4e49c one multiclass now defines all 8 variants of binary-scalar-sse-fp operations. 
llvm-svn: 30814
 2006-10-07 20:55:57 +00:00
Chris Lattner 6eaee2c8e3 Switch ADD/MUL/DIV/SUB scalarsse fp ops to a multiclass 
llvm-svn: 30813
 2006-10-07 20:35:44 +00:00
Chris Lattner c8c6441821 Random acts of shrinkage 
llvm-svn: 30812
 2006-10-07 19:49:05 +00:00
Chris Lattner b5df7e554d Convert pand/por/pxor to use multiclass 
llvm-svn: 30811
 2006-10-07 19:37:30 +00:00
Chris Lattner 6138cba5f1 Convert some more instructions over to use a new multiclass. 
Fix a bug where the asmstring for PSUBQrm was wrong.

llvm-svn: 30810
 2006-10-07 19:34:33 +00:00
Chris Lattner 662ba43f08 Fix a bug where PADDQrm printed paddd instead of paddq. 
llvm-svn: 30809
 2006-10-07 19:15:46 +00:00
Chris Lattner 29c62a3c88 Add multiclass for SSE2 instructions that correspond to simple binops. 
llvm-svn: 30808
 2006-10-07 19:14:49 +00:00
Chris Lattner e0928d9d7b rename: 
PDI_binop_rm -> PDI_binop_rm_int
  PDI_binop_rmi -> PDI_binop_rmi_int

to make it clear that these are for use with intrinsics.

llvm-svn: 30807
 2006-10-07 19:02:31 +00:00
Chris Lattner 489b63089d Convert saturating PADD/PSUB's to use a multiclass 
llvm-svn: 30806
 2006-10-07 18:48:46 +00:00
Chris Lattner fa2ce8824d Convert PAVG*, PMADDWD, and PMUL* to use multiclasses. 
llvm-svn: 30805
 2006-10-07 18:39:00 +00:00
Chris Lattner cab92e4c0c Fix typo in packsswb instr definition, where the load had the wrong type. 
This allows us to use the multiclass for other packs.

llvm-svn: 30804
 2006-10-07 18:23:58 +00:00
Chris Lattner e746a9cd6a handle pmin/pmax with multiclasses 
llvm-svn: 30800
 2006-10-07 07:49:33 +00:00
Chris Lattner b14e6a0f8c simplify pack and shift intrinsics with multiclasses 
llvm-svn: 30797
 2006-10-07 07:06:17 +00:00
Chris Lattner 521fc4e33f Use a multiclass to simplify 'SSE2 Integer comparison' 
llvm-svn: 30796
 2006-10-07 06:47:08 +00:00
Chris Lattner c6138cec61 move class defns close to uses to make it easier to read 
llvm-svn: 30795
 2006-10-07 06:33:36 +00:00
Chris Lattner 87e692323c simplify horizontal op definitions 
llvm-svn: 30794
 2006-10-07 06:31:41 +00:00
Chris Lattner 3e9fc37458 remove more unneeded type info 
llvm-svn: 30793
 2006-10-07 06:27:03 +00:00
Chris Lattner 807be0a715 remove unneeded definitions and type info 
llvm-svn: 30792
 2006-10-07 06:19:41 +00:00
Chris Lattner 5b1358a8eb remove some unneeded type info 
llvm-svn: 30791
 2006-10-07 06:17:43 +00:00
Chris Lattner 3414c022af simplify patterns by merging in operand info 
llvm-svn: 30790
 2006-10-07 05:50:25 +00:00
Chris Lattner ca21ce5f08 Factor operands into packed unary classes 
llvm-svn: 30789
 2006-10-07 05:47:20 +00:00
Chris Lattner 0052c3ff5b remove dead/duplicate instructions 
llvm-svn: 30788
 2006-10-07 05:41:52 +00:00
Chris Lattner 904c6e9c92 Pull operand info up into parent class for scalar sse intrinsics. 
llvm-svn: 30787
 2006-10-07 05:26:13 +00:00
Chris Lattner e698c90ee9 convert the sole sd unary intrinsic to a multiclass for consistency 
llvm-svn: 30786
 2006-10-07 05:19:31 +00:00
Chris Lattner 2bb2f050f5 pull operand string into the multiclass 
llvm-svn: 30785
 2006-10-07 05:13:26 +00:00
Chris Lattner 069679c7b6 Remove RSQRTSS[rm] RCPSS[rm], which are dead. 
Introduce SS_IntUnary, a multiclass to replace SS_Int[rm].

llvm-svn: 30784
 2006-10-07 05:09:48 +00:00
Chris Lattner f13a7b376c eliminate redundancy 
llvm-svn: 30783
 2006-10-07 04:52:09 +00:00
Evan Cheng a36e6cf44f These don't have immediate operands. 
llvm-svn: 30694
 2006-10-03 06:55:11 +00:00
Evan Cheng 4259a0f654 X86ISD::CMP now produces a chain as well as a flag. Make that the chain 
operand of a conditional branch to allow load folding into CMP / TEST
instructions.

llvm-svn: 30241
 2006-09-11 02:19:56 +00:00
Evan Cheng 17c28b2e0e JIT encoding bug. 
llvm-svn: 30112
 2006-09-05 05:59:25 +00:00
Evan Cheng 66ed41cac1 Can't commute shufps. The high / low parts elements come from different vectors. 
llvm-svn: 29275
 2006-07-25 20:25:40 +00:00
Evan Cheng 5987cfb7b1 X86 target specific DAG combine: turn build_vector (load x), (load x+4), 
(load x+8), (load x+12), <0, 1, 2, 3> to a single 128-bit load (aligned and
unaligned).

e.g.

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

_test:
        movups 4(%esp), %xmm0
        ret

llvm-svn: 29042
 2006-07-07 08:33:52 +00:00
Evan Cheng 390922f979 Should just use xorps to clear XMM registers for all data types. pxor is also one byte longer. 
llvm-svn: 28984
 2006-06-29 18:04:54 +00:00
Evan Cheng fc8cdda070 Always use xorps to clear XMM registers. 
llvm-svn: 28979
 2006-06-29 00:34:23 +00:00
Chris Lattner dbec49d574 Remove some ugly now-redundant casts. 
llvm-svn: 28864
 2006-06-20 00:25:29 +00:00
Chris Lattner 55594634d7 Fix some mismatched type constraints 
llvm-svn: 28862
 2006-06-20 00:12:37 +00:00
Evan Cheng cd58e9d8b9 Minor clean up. 
llvm-svn: 28860
 2006-06-19 19:25:30 +00:00
Evan Cheng de7156f12c Type of vector extract / insert index operand should be iPTR. 
llvm-svn: 28796
 2006-06-15 08:14:54 +00:00
Evan Cheng 25e44e008d Rename instructions for consistency sake. 
llvm-svn: 28594
 2006-05-31 19:00:07 +00:00
Evan Cheng 8abf45e22d Select vector_shuffle v1, undef <2, 3, ?, ?> to MOVHLPS. 
llvm-svn: 28582
 2006-05-31 00:51:37 +00:00
Evan Cheng 57399704b3 MAXP{D|S} and MINP{D|S} are commutable. 
llvm-svn: 28578
 2006-05-30 23:47:30 +00:00
Evan Cheng c0f90bef47 Commute shufps / shufpd. 
llvm-svn: 28577
 2006-05-30 23:34:30 +00:00
Evan Cheng 66f849bd7b Allow shufps x, x, mask to be converted to pshufd x, mask to save a move. 
llvm-svn: 28565
 2006-05-30 20:26:50 +00:00
Evan Cheng 9fee442e63 X86 integer register classes naming changes. Make them consistent with FP, vector classes. 
llvm-svn: 28324
 2006-05-16 07:21:53 +00:00
Chris Lattner 44a73e9fa5 Teach the code generator to use cvtss2sd as extload f32 -> f64 
llvm-svn: 28131
 2006-05-05 21:35:18 +00:00
Evan Cheng 8b1cde2bbe Use movsd to shuffle in the lowest two elements of a v4f32 / v4i32 vector when 
movlps cannot be used (e.g. when load from m64 has multiple uses).

llvm-svn: 28089
 2006-05-03 20:32:03 +00:00
Evan Cheng 4cc3e0b05f Fix a typo. 
llvm-svn: 27968
 2006-04-25 17:48:41 +00:00
Evan Cheng fb46b2bf5d Explicitly specify result type for def : Pat<> patterns (if it produces a vector 
result). Otherwise tblgen will pick the default (v16i8 for 128-bit vector).

llvm-svn: 27965
 2006-04-25 00:50:01 +00:00
Evan Cheng 25b09295f8 Added X86 SSE2 intrinsics which can be represented as vector_shuffles. This is 
a temporary workaround for the 2-wide vector_shuffle problem (i.e. its mask
would have type v2i32 which is not legal).

llvm-svn: 27964
 2006-04-24 23:34:56 +00:00
Evan Cheng 63bd4d3730 Some missing movlps, movhps, movlpd, and movhpd patterns. 
llvm-svn: 27960
 2006-04-24 21:58:20 +00:00
Evan Cheng e8b5180044 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

llvm-svn: 27923
 2006-04-21 01:05:10 +00:00
Evan Cheng 6d5297dac3 Prefer {p}unpack* and mov*dup over {p}shuf* as well. 
llvm-svn: 27844
 2006-04-19 21:15:24 +00:00
Evan Cheng b416a25174 - Renamed AddedCost to AddedComplexity. 
- Added more movhlps and movlhps patterns.

llvm-svn: 27842
 2006-04-19 20:37:34 +00:00
Evan Cheng cc7abc6c38 More mov{h|l}p{d|s} patterns. 
llvm-svn: 27836
 2006-04-19 18:20:17 +00:00
Evan Cheng aeb09ccdd3 - 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.

llvm-svn: 27835
 2006-04-19 18:11:52 +00:00
Evan Cheng 3823aa1d0f - PEXTRW cannot take a memory location as its first source operand. 
- PINSRWrmi encoding bug.

llvm-svn: 27818
 2006-04-18 21:59:43 +00:00
Evan Cheng a179ea631d Name change for clarity sake 
llvm-svn: 27816
 2006-04-18 21:55:35 +00:00
Evan Cheng d799d680f4 Name change for clarity sake 
llvm-svn: 27814
 2006-04-18 21:29:50 +00:00
Evan Cheng 0ee281f37c Left a pattern out 
llvm-svn: 27813
 2006-04-18 21:29:08 +00:00
Evan Cheng e2d25a1a50 Fixed an encoding bug: movd from XMM to R32. 
llvm-svn: 27807
 2006-04-18 18:19:00 +00:00
Evan Cheng 5421206c4b Use movss to insert_vector_elt(v, s, 0). 
llvm-svn: 27782
 2006-04-17 22:45:49 +00:00
Evan Cheng 22c06f054b Encoding bug 
llvm-svn: 27773
 2006-04-17 21:33:57 +00:00
Evan Cheng 5112b5c544 Errors in patterns preventing load folding 
llvm-svn: 27762
 2006-04-17 18:05:01 +00:00
Evan Cheng 20712deecb movduprm, movshduprm bugs 
llvm-svn: 27734
 2006-04-16 18:11:28 +00:00
Evan Cheng 3064f9aaa6 Encoding bugs 
llvm-svn: 27733
 2006-04-16 07:02:22 +00:00
Evan Cheng 8f1d801389 More encoding bugs 
llvm-svn: 27722
 2006-04-15 06:10:09 +00:00
Evan Cheng 91944e8699 pslldrm, psrawrm, etc. encoding bug 
llvm-svn: 27721
 2006-04-15 05:59:08 +00:00
Evan Cheng 1220b31a31 hsubp{s|d} encoding bug 
llvm-svn: 27720
 2006-04-15 05:52:42 +00:00
Evan Cheng 6222cf2a36 Silly bug 
llvm-svn: 27719
 2006-04-15 05:37:34 +00:00
Evan Cheng 00a5b3d9d3 Some clean up 
llvm-svn: 27715
 2006-04-14 23:32:40 +00:00
Evan Cheng 5d247f81c1 Last few SSE3 intrinsics. 
llvm-svn: 27711
 2006-04-14 21:59:03 +00:00
Evan Cheng 3bd605397b Misc. SSE2 intrinsics: clflush, lfench, mfence 
llvm-svn: 27699
 2006-04-14 07:43:12 +00:00
Evan Cheng eb0063a34f pcmpeq* and pcmpgt* intrinsics. 
llvm-svn: 27685
 2006-04-14 01:39:53 +00:00
Evan Cheng 16287444ff psll*, psrl*, and psra* intrinsics. 
llvm-svn: 27684
 2006-04-14 00:14:05 +00:00
Evan Cheng a84319719c Doh. PANDrm, etc. are not commutable. 
llvm-svn: 27668
 2006-04-13 18:11:28 +00:00
Evan Cheng ed3996743f psad, pmax, pmin intrinsics. 
llvm-svn: 27647
 2006-04-13 06:11:45 +00:00
Evan Cheng 58dad55959 Various SSE2 packed integer intrinsics: pmulhuw, pavgw, etc. 
llvm-svn: 27645
 2006-04-13 05:24:54 +00:00
Evan Cheng b3fe00bdc6 padds{b|w}, paddus{b|w}, psubs{b|w}, psubus{b|w} intrinsics. 
llvm-svn: 27639
 2006-04-13 00:43:35 +00:00
Evan Cheng 0aab735a1a Naming inconsistency. 
llvm-svn: 27638
 2006-04-13 00:00:23 +00:00
Evan Cheng c88afc36a9 SSE / SSE2 conversion intrinsics. 
llvm-svn: 27637
 2006-04-12 23:42:44 +00:00
Evan Cheng 92232307d0 All "integer" logical ops (pand, por, pxor) are now promoted to v2i64. 
Clean up and fix various logical ops issues.

llvm-svn: 27633
 2006-04-12 21:21:57 +00:00
Evan Cheng e2157c6e41 Promote v4i32, v8i16, v16i8 load to v2i64 load. 
llvm-svn: 27612
 2006-04-12 17:12:36 +00:00
Evan Cheng 29be057d92 Various SSE2 conversion intrinsics 
llvm-svn: 27603
 2006-04-12 05:20:24 +00:00
Evan Cheng 70c74a3ced Added __builtin_ia32_storelv4si, __builtin_ia32_movqv4si, 
__builtin_ia32_loadlv4si, __builtin_ia32_loaddqu, __builtin_ia32_storedqu.

llvm-svn: 27599
 2006-04-11 22:28:25 +00:00
Evan Cheng 6b60357f4a gcc lower SSE prefetch into generic prefetch intrinsic. Need to add support 
later.

llvm-svn: 27591
 2006-04-11 18:04:57 +00:00
Evan Cheng 6ea715af28 Misc. intrinsics. 
llvm-svn: 27590
 2006-04-11 17:35:57 +00:00
Evan Cheng 09a956271a movnt* and maskmovdqu intrinsics 
llvm-svn: 27587
 2006-04-11 06:57:30 +00:00
Evan Cheng 12ba3e23d0 Added support for _mm_move_ss and _mm_move_sd. 
llvm-svn: 27575
 2006-04-11 00:19:04 +00:00
Evan Cheng f8ac02283c Remove some bogus patterns; clean up. 
llvm-svn: 27569
 2006-04-10 22:35:16 +00:00
Evan Cheng 76112c3cb8 Added some missing shuffle patterns. 
llvm-svn: 27564
 2006-04-10 21:42:19 +00:00
Evan Cheng 395fa3d2a6 movups / movupd 
llvm-svn: 27562
 2006-04-10 21:11:06 +00:00
Evan Cheng 617a6a812e Conditional move of vector types. 
llvm-svn: 27556
 2006-04-10 07:23:14 +00:00
Evan Cheng 0df9c9f57d ldmxcsr and stmxcsr. 
llvm-svn: 27506
 2006-04-08 00:47:44 +00:00
Evan Cheng aa18a52545 Added patterns for MOVHPSmr and MOVLPSmr. 
llvm-svn: 27497
 2006-04-07 21:20:58 +00:00
Evan Cheng d8e1a01be6 A MOVPS2SSmr, i.e. _mm_store_ss, encoding bug. 
Also MOVPDI2DIrr.

llvm-svn: 27476
 2006-04-06 23:53:29 +00:00
Evan Cheng c995b45f67 - 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.

llvm-svn: 27474
 2006-04-06 23:23:56 +00:00
Evan Cheng 695e45c252 POR encoded as PAND, yikes. 
llvm-svn: 27446
 2006-04-06 01:49:20 +00:00
Evan Cheng 780382946e Support for comi / ucomi intrinsics. 
llvm-svn: 27444
 2006-04-05 23:38:46 +00:00
Evan Cheng f3b52c84ea 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.

llvm-svn: 27437
 2006-04-05 07:20:06 +00:00
Evan Cheng 011c23d9d3 Added pslldq and psrldq. 
llvm-svn: 27412
 2006-04-04 21:49:39 +00:00
Evan Cheng 8f3b6b8d8a Minor fixes + naming changes. 
llvm-svn: 27410
 2006-04-04 19:12:30 +00:00
Evan Cheng 802b35c339 PSHUF* encoding bugs. 
llvm-svn: 27405
 2006-04-04 18:40:36 +00:00
Evan Cheng e91e3bd874 cmpps / cmppd encoding bug 
llvm-svn: 27393
 2006-04-04 03:04:07 +00:00
Evan Cheng dd2eb27d6d Compact some intrinsic definitions. 
llvm-svn: 27388
 2006-04-04 00:10:53 +00:00
Evan Cheng 0ef83c83e1 Some SSE1 intrinsics: min, max, sqrt, etc. 
llvm-svn: 27384
 2006-04-03 23:49:17 +00:00
Evan Cheng b64827e662 Use movlpd to: store lower f64 extracted from v2f64. 
Use movhpd to: store upper f64 extracted from v2f64.

llvm-svn: 27382
 2006-04-03 22:30:54 +00:00
Evan Cheng ebf1006d16 - More efficient extract_vector_elt with shuffle and movss, movsd, movd, etc. 
- Some bug fixes and naming inconsistency fixes.

llvm-svn: 27377
 2006-04-03 20:53:28 +00:00
Evan Cheng 5fd7c69473 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.

llvm-svn: 27314
 2006-03-31 21:55:24 +00:00
Evan Cheng 747e29ef0b Added support for SSE3 horizontal ops: haddp{s|d} and hsub{s|d}. 
llvm-svn: 27310
 2006-03-31 21:29:33 +00:00
Evan Cheng cbffa4656b Add support to use pextrw and pinsrw to extract and insert a word element 
from a 128-bit vector.

llvm-svn: 27304
 2006-03-31 19:22:53 +00:00
Evan Cheng 7e2ff11a42 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.

llvm-svn: 27259
 2006-03-30 19:54:57 +00:00
Evan Cheng dd487d865b More logical ops patterns 
llvm-svn: 27257
 2006-03-30 07:33:32 +00:00
Evan Cheng c58ef7deeb Add support for _mm_cmp{cc}_ss and _mm_cmp{cc}_ps intrinsics 
llvm-svn: 27256
 2006-03-30 06:21:22 +00:00
Evan Cheng 593310016d Add 128-bit pmovmskb intrinsic support. 
llvm-svn: 27255
 2006-03-30 00:33:26 +00:00
Evan Cheng c5cf9bba05 Change SSE pack operation definitions to fit what the intrinsics expected. 
For example, packsswb actually creates a v16i8 from a pair of v8i16. But since
the intrinsic specification forces the output type to match the operands.

llvm-svn: 27254
 2006-03-29 23:53:14 +00:00
Evan Cheng b7fedffc78 - Added some SSE2 128-bit packed integer ops. 
- Added SSE2 128-bit integer pack with signed saturation ops.
- Added pshufhw and pshuflw ops.

llvm-svn: 27252
 2006-03-29 23:07:14 +00:00
Evan Cheng acc336475e Need to special case splat after all. Make the second operand of splat 
vector_shuffle undef.

llvm-svn: 27250
 2006-03-29 19:02:40 +00:00
Evan Cheng 3cf95747c7 Floating point logical operation patterns should match bit_convert. Or else 
integer vector logical operations would match andp{s|d} instead of pand.

llvm-svn: 27248
 2006-03-29 18:47:40 +00:00
Evan Cheng 500ec16578 - More shuffle related bug fixes. 
- Whenever possible use ops of the right packed types for vector shuffles /
  splats.

llvm-svn: 27246
 2006-03-29 03:04:49 +00:00
Evan Cheng da59b0d2a8 - Only use pshufd for v4i32 vector shuffles. 
- Other shuffle related fixes.

llvm-svn: 27244
 2006-03-29 01:30:51 +00:00
Evan Cheng 38b34296d0 Added aliases to scalar SSE instructions, e.g. addss, to match x86 intrinsics. 
The source operands type are v4sf with upper bits passes through.
Added matching code for these.

llvm-svn: 27240
 2006-03-28 23:51:43 +00:00
Evan Cheng be2d9a0e99 movlps and movlpd should be modeled as two address code. 
llvm-svn: 27221
 2006-03-28 07:01:28 +00:00
Evan Cheng 4e7374ff8a Typo 
llvm-svn: 27219
 2006-03-28 06:53:49 +00:00
Evan Cheng 1a194a5264 * Prefer using operation of matching types. e.g unpcklpd rather than movlhps. 
* Bug fixes.

llvm-svn: 27218
 2006-03-28 06:50:32 +00:00
Evan Cheng 2bc3280659 - Clean up / consoladate various shuffle masks. 
- Some misc. bug fixes.
- Use MOVHPDrm to load from m64 to upper half of a XMM register.

llvm-svn: 27210
 2006-03-28 02:43:26 +00:00
Evan Cheng 5df75889db Model unpack lower and interleave as vector_shuffle so we can lower the 
intrinsics as such.

llvm-svn: 27200
 2006-03-28 00:39:58 +00:00
Chris Lattner 018e17c8de unbreak the build 
llvm-svn: 27174
 2006-03-27 16:52:45 +00:00
Evan Cheng 9b9cc4fb39 Use pcmpeq to generate vector of all ones. 
llvm-svn: 27167
 2006-03-27 07:00:16 +00:00
Evan Cheng ed6184aef2 Remove X86:isZeroVector, use ISD::isBuildVectorAllZeros instead; some fixes / cleanups 
llvm-svn: 27150
 2006-03-26 09:53:12 +00:00
Evan Cheng 2bc0941e2a Build arbitrary vector with more than 2 distinct scalar elements with a 
series of unpack and interleave ops.

llvm-svn: 27119
 2006-03-25 09:37:23 +00:00
Evan Cheng 79e500ec74 Added SSE cachebility ops 
llvm-svn: 27103
 2006-03-25 06:03:26 +00:00
Evan Cheng 6f7d31ea50 Added 128-bit packed integer subtraction. 
llvm-svn: 27096
 2006-03-25 01:33:37 +00:00
Evan Cheng 980c4d5b46 Added CVTSS2SI. 
llvm-svn: 27094
 2006-03-25 01:00:18 +00:00
Evan Cheng e7ee6a5e32 Support for scalar to vector with zero extension. 
llvm-svn: 27091
 2006-03-24 23:15:12 +00:00
Evan Cheng 2f0277bf48 Added LDMXCSR 
llvm-svn: 27087
 2006-03-24 22:28:37 +00:00
Chris Lattner 97599f1211 plug the intrinsics into the patterns for movmsk* 
llvm-svn: 27083
 2006-03-24 21:49:18 +00:00
Evan Cheng 082c8785ef Handle BUILD_VECTOR with all zero elements. 
llvm-svn: 27056
 2006-03-24 07:29:27 +00:00
Evan Cheng 2595a687da More efficient v2f64 shuffle using movlhps, movhlps, unpckhpd, and unpcklpd. 
llvm-svn: 27040
 2006-03-24 02:58:06 +00:00
Evan Cheng d27fb3e85e Handle more shuffle cases with SHUFP* instructions. 
llvm-svn: 27024
 2006-03-24 01:18:28 +00:00
Evan Cheng 82ed4a42f9 Following icc's lead: use movdqa to load / store 128-bit integer vectors 
llvm-svn: 26980
 2006-03-23 07:44:07 +00:00
Evan Cheng 7055878170 Add v4i32 <-> v4f32 bitconvert patterns. 
llvm-svn: 26969
 2006-03-23 02:36:37 +00:00
Evan Cheng b9b0550dc6 Add 128-bit integer vector load and add (for testing). 
llvm-svn: 26967
 2006-03-23 01:57:24 +00:00
Evan Cheng ed794cd27b SHUFP* are two address code. 
llvm-svn: 26959
 2006-03-22 20:08:18 +00:00
Evan Cheng d4e1557941 - Supposely movlhps is faster / better than unpcklpd. 
- Don't forget pshufd is only available with sse2.

llvm-svn: 26956
 2006-03-22 19:16:21 +00:00
Evan Cheng 68ad48bd1a - Implement X86ISelLowering::isShuffleMaskLegal(). We currently only support 
splat and PSHUFD cases.
- Clean up shuffle / splat matching code.

llvm-svn: 26954
 2006-03-22 18:59:22 +00:00
Evan Cheng 8fdbdf20cd - VECTOR_SHUFFLE of v4i32 / v4f32 with undef second vector always matches 
PSHUFD. We can make permutes entries which point to the undef pointing
  anything we want.
- Change some names to appease Chris.

llvm-svn: 26951
 2006-03-22 08:01:21 +00:00
Evan Cheng 3617caf526 Fix PSHUF* and SHUF* jit code emission problems 
llvm-svn: 26949
 2006-03-22 07:10:28 +00:00
Evan Cheng d097e67544 Some splat and shuffle support. 
llvm-svn: 26940
 2006-03-22 02:53:00 +00:00
Evan Cheng d5e905d762 - Use movaps to store 128-bit vector integers. 
- Each scalar to vector v8i16 and v16i8 is a any_extend followed by a movd.

llvm-svn: 26932
 2006-03-21 23:01:21 +00:00
Evan Cheng 1208d9179a - Remove scalar to vector pseudo ops. They are just wrong. 
- Handle FR32 to VR128:v4f32 and FR64 to VR128:v2f64 with aliases of MOVAPS
and MOVAPD. Mark them as move instructions and *hope* they will be deleted.

llvm-svn: 26919
 2006-03-21 07:09:35 +00:00
Evan Cheng e4d1416239 x86 ISD::SCALAR_TO_VECTOR support. 
llvm-svn: 26911
 2006-03-21 00:33:35 +00:00
Evan Cheng e6448448c2 Move a few things around. 
llvm-svn: 26893
 2006-03-20 06:04:52 +00:00
Evan Cheng f7c2e3628b Vector undef's 
llvm-svn: 26870
 2006-03-19 09:38:54 +00:00
Evan Cheng 9bf978dc20 Use the generic vector register classes VR64 / VR128 rather than V4F32, 
V8I16, etc.

llvm-svn: 26838
 2006-03-18 01:23:20 +00:00
Evan Cheng 4f674921d6 Move some pattern fragments to the right files. 
llvm-svn: 26831
 2006-03-17 19:55:52 +00:00
Evan Cheng 9e252e3bcf Added MMX, SSE1, and SSE2 vector instructions and some simple patterns. 
Fixed some existing bugs (wrong predicates, prefixes) at the same time.

llvm-svn: 26310
 2006-02-22 02:26:30 +00:00
Evan Cheng d58478161f One more round of reorg so sabre doesn't freak out. :-) 
llvm-svn: 26303
 2006-02-21 20:00:20 +00:00
Evan Cheng 6fc1162855 A big more cleaning up. 
llvm-svn: 26302
 2006-02-21 19:30:30 +00:00
Evan Cheng 8711b6bff3 Moving things to their proper places. 
llvm-svn: 26301
 2006-02-21 19:26:52 +00:00
Evan Cheng 6e595b9fd8 Split instruction info into multiple files, one for each of x87, MMX, and SSE. 
llvm-svn: 26300
 2006-02-21 19:13:53 +00:00