summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--llvm/lib/Target/X86/InstPrinter/X86InstComments.cpp8
-rw-r--r--llvm/lib/Target/X86/X86InstrAVX512.td22
-rw-r--r--llvm/lib/Target/X86/X86InstrInfo.cpp6
-rw-r--r--llvm/lib/Target/X86/X86InstrSSE.td38
-rw-r--r--llvm/test/CodeGen/X86/sad.ll6
-rw-r--r--llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll4
-rw-r--r--llvm/test/CodeGen/X86/vector-shuffle-v1.ll2
7 files changed, 41 insertions, 45 deletions
diff --git a/llvm/lib/Target/X86/InstPrinter/X86InstComments.cpp b/llvm/lib/Target/X86/InstPrinter/X86InstComments.cpp
index bc0c5351363..69e82343907 100644
--- a/llvm/lib/Target/X86/InstPrinter/X86InstComments.cpp
+++ b/llvm/lib/Target/X86/InstPrinter/X86InstComments.cpp
@@ -342,14 +342,10 @@ bool llvm::EmitAnyX86InstComments(const MCInst *MI, raw_ostream &OS,
ShuffleMask);
break;
- case X86::PALIGNR128rr:
- case X86::VPALIGNR128rr:
- case X86::VPALIGNR256rr:
+ CASE_SHUF(PALIGNR, rri)
Src1Name = getRegName(MI->getOperand(2).getReg());
// FALL THROUGH.
- case X86::PALIGNR128rm:
- case X86::VPALIGNR128rm:
- case X86::VPALIGNR256rm:
+ CASE_SHUF(PALIGNR, rmi)
Src2Name = getRegName(MI->getOperand(1).getReg());
DestName = getRegName(MI->getOperand(0).getReg());
if (MI->getOperand(NumOperands - 1).isImm())
diff --git a/llvm/lib/Target/X86/X86InstrAVX512.td b/llvm/lib/Target/X86/X86InstrAVX512.td
index 0ab57db9403..83bfa36cfb4 100644
--- a/llvm/lib/Target/X86/X86InstrAVX512.td
+++ b/llvm/lib/Target/X86/X86InstrAVX512.td
@@ -7224,7 +7224,7 @@ defm VALIGND: avx512_valign<"valignd", avx512vl_i32_info>,
defm VALIGNQ: avx512_valign<"valignq", avx512vl_i64_info>,
EVEX_CD8<64, CD8VF>, VEX_W;
-multiclass avx512_vpalign_lowering<X86VectorVTInfo _ , list<Predicate> p>{
+multiclass avx512_vpalignr_lowering<X86VectorVTInfo _ , list<Predicate> p>{
let Predicates = p in
def NAME#_.VTName#rri:
Pat<(_.VT (X86PAlignr _.RC:$src1, _.RC:$src2, (i8 imm:$imm))),
@@ -7232,18 +7232,18 @@ multiclass avx512_vpalign_lowering<X86VectorVTInfo _ , list<Predicate> p>{
_.RC:$src1, _.RC:$src2, imm:$imm)>;
}
-multiclass avx512_vpalign_lowering_common<AVX512VLVectorVTInfo _>:
- avx512_vpalign_lowering<_.info512, [HasBWI]>,
- avx512_vpalign_lowering<_.info128, [HasBWI, HasVLX]>,
- avx512_vpalign_lowering<_.info256, [HasBWI, HasVLX]>;
+multiclass avx512_vpalignr_lowering_common<AVX512VLVectorVTInfo _>:
+ avx512_vpalignr_lowering<_.info512, [HasBWI]>,
+ avx512_vpalignr_lowering<_.info128, [HasBWI, HasVLX]>,
+ avx512_vpalignr_lowering<_.info256, [HasBWI, HasVLX]>;
-defm VPALIGN: avx512_common_3Op_rm_imm8<0x0F, X86PAlignr, "vpalignr" ,
+defm VPALIGNR: avx512_common_3Op_rm_imm8<0x0F, X86PAlignr, "vpalignr" ,
avx512vl_i8_info, avx512vl_i8_info>,
- avx512_vpalign_lowering_common<avx512vl_i16_info>,
- avx512_vpalign_lowering_common<avx512vl_i32_info>,
- avx512_vpalign_lowering_common<avx512vl_f32_info>,
- avx512_vpalign_lowering_common<avx512vl_i64_info>,
- avx512_vpalign_lowering_common<avx512vl_f64_info>,
+ avx512_vpalignr_lowering_common<avx512vl_i16_info>,
+ avx512_vpalignr_lowering_common<avx512vl_i32_info>,
+ avx512_vpalignr_lowering_common<avx512vl_f32_info>,
+ avx512_vpalignr_lowering_common<avx512vl_i64_info>,
+ avx512_vpalignr_lowering_common<avx512vl_f64_info>,
EVEX_CD8<8, CD8VF>;
defm VDBPSADBW: avx512_common_3Op_rm_imm8<0x42, X86dbpsadbw, "vdbpsadbw" ,
diff --git a/llvm/lib/Target/X86/X86InstrInfo.cpp b/llvm/lib/Target/X86/X86InstrInfo.cpp
index 422c9de9bb3..32ab08724db 100644
--- a/llvm/lib/Target/X86/X86InstrInfo.cpp
+++ b/llvm/lib/Target/X86/X86InstrInfo.cpp
@@ -1029,7 +1029,7 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
{ X86::PADDUSBrr, X86::PADDUSBrm, TB_ALIGN_16 },
{ X86::PADDUSWrr, X86::PADDUSWrm, TB_ALIGN_16 },
{ X86::PADDWrr, X86::PADDWrm, TB_ALIGN_16 },
- { X86::PALIGNR128rr, X86::PALIGNR128rm, TB_ALIGN_16 },
+ { X86::PALIGNRrri, X86::PALIGNRrmi, TB_ALIGN_16 },
{ X86::PANDNrr, X86::PANDNrm, TB_ALIGN_16 },
{ X86::PANDrr, X86::PANDrm, TB_ALIGN_16 },
{ X86::PAVGBrr, X86::PAVGBrm, TB_ALIGN_16 },
@@ -1326,7 +1326,7 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
{ X86::VPADDUSBrr, X86::VPADDUSBrm, 0 },
{ X86::VPADDUSWrr, X86::VPADDUSWrm, 0 },
{ X86::VPADDWrr, X86::VPADDWrm, 0 },
- { X86::VPALIGNR128rr, X86::VPALIGNR128rm, 0 },
+ { X86::VPALIGNRrri, X86::VPALIGNRrmi, 0 },
{ X86::VPANDNrr, X86::VPANDNrm, 0 },
{ X86::VPANDrr, X86::VPANDrm, 0 },
{ X86::VPAVGBrr, X86::VPAVGBrm, 0 },
@@ -1482,7 +1482,7 @@ X86InstrInfo::X86InstrInfo(X86Subtarget &STI)
{ X86::VPADDUSBYrr, X86::VPADDUSBYrm, 0 },
{ X86::VPADDUSWYrr, X86::VPADDUSWYrm, 0 },
{ X86::VPADDWYrr, X86::VPADDWYrm, 0 },
- { X86::VPALIGNR256rr, X86::VPALIGNR256rm, 0 },
+ { X86::VPALIGNRYrri, X86::VPALIGNRYrmi, 0 },
{ X86::VPANDNYrr, X86::VPANDNYrm, 0 },
{ X86::VPANDYrr, X86::VPANDYrm, 0 },
{ X86::VPAVGBYrr, X86::VPAVGBYrm, 0 },
diff --git a/llvm/lib/Target/X86/X86InstrSSE.td b/llvm/lib/Target/X86/X86InstrSSE.td
index 3ba72f788c4..3f1f4e71349 100644
--- a/llvm/lib/Target/X86/X86InstrSSE.td
+++ b/llvm/lib/Target/X86/X86InstrSSE.td
@@ -5674,7 +5674,7 @@ defm PMULHRSW : SS3I_binop_rm_int<0x0B, "pmulhrsw",
multiclass ssse3_palignr<string asm, bit Is2Addr = 1> {
let hasSideEffects = 0 in {
- def R128rr : SS3AI<0x0F, MRMSrcReg, (outs VR128:$dst),
+ def rri : SS3AI<0x0F, MRMSrcReg, (outs VR128:$dst),
(ins VR128:$src1, VR128:$src2, u8imm:$src3),
!if(Is2Addr,
!strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
@@ -5682,7 +5682,7 @@ multiclass ssse3_palignr<string asm, bit Is2Addr = 1> {
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
[], IIC_SSE_PALIGNRR>, Sched<[WriteShuffle]>;
let mayLoad = 1 in
- def R128rm : SS3AI<0x0F, MRMSrcMem, (outs VR128:$dst),
+ def rmi : SS3AI<0x0F, MRMSrcMem, (outs VR128:$dst),
(ins VR128:$src1, i128mem:$src2, u8imm:$src3),
!if(Is2Addr,
!strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
@@ -5694,13 +5694,13 @@ multiclass ssse3_palignr<string asm, bit Is2Addr = 1> {
multiclass ssse3_palignr_y<string asm, bit Is2Addr = 1> {
let hasSideEffects = 0 in {
- def R256rr : SS3AI<0x0F, MRMSrcReg, (outs VR256:$dst),
+ def Yrri : SS3AI<0x0F, MRMSrcReg, (outs VR256:$dst),
(ins VR256:$src1, VR256:$src2, u8imm:$src3),
!strconcat(asm,
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
[]>, Sched<[WriteShuffle]>;
let mayLoad = 1 in
- def R256rm : SS3AI<0x0F, MRMSrcMem, (outs VR256:$dst),
+ def Yrmi : SS3AI<0x0F, MRMSrcMem, (outs VR256:$dst),
(ins VR256:$src1, i256mem:$src2, u8imm:$src3),
!strconcat(asm,
"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
@@ -5709,43 +5709,43 @@ multiclass ssse3_palignr_y<string asm, bit Is2Addr = 1> {
}
let Predicates = [HasAVX] in
- defm VPALIGN : ssse3_palignr<"vpalignr", 0>, VEX_4V;
+ defm VPALIGNR : ssse3_palignr<"vpalignr", 0>, VEX_4V;
let Predicates = [HasAVX2] in
- defm VPALIGN : ssse3_palignr_y<"vpalignr", 0>, VEX_4V, VEX_L;
+ defm VPALIGNR : ssse3_palignr_y<"vpalignr", 0>, VEX_4V, VEX_L;
let Constraints = "$src1 = $dst", Predicates = [UseSSSE3] in
- defm PALIGN : ssse3_palignr<"palignr">;
+ defm PALIGNR : ssse3_palignr<"palignr">;
let Predicates = [HasAVX2, NoVLX_Or_NoBWI] in {
def : Pat<(v8i32 (X86PAlignr VR256:$src1, VR256:$src2, (i8 imm:$imm))),
- (VPALIGNR256rr VR256:$src1, VR256:$src2, imm:$imm)>;
+ (VPALIGNRYrri VR256:$src1, VR256:$src2, imm:$imm)>;
def : Pat<(v8f32 (X86PAlignr VR256:$src1, VR256:$src2, (i8 imm:$imm))),
- (VPALIGNR256rr VR256:$src1, VR256:$src2, imm:$imm)>;
+ (VPALIGNRYrri VR256:$src1, VR256:$src2, imm:$imm)>;
def : Pat<(v16i16 (X86PAlignr VR256:$src1, VR256:$src2, (i8 imm:$imm))),
- (VPALIGNR256rr VR256:$src1, VR256:$src2, imm:$imm)>;
+ (VPALIGNRYrri VR256:$src1, VR256:$src2, imm:$imm)>;
def : Pat<(v32i8 (X86PAlignr VR256:$src1, VR256:$src2, (i8 imm:$imm))),
- (VPALIGNR256rr VR256:$src1, VR256:$src2, imm:$imm)>;
+ (VPALIGNRYrri VR256:$src1, VR256:$src2, imm:$imm)>;
}
let Predicates = [HasAVX, NoVLX_Or_NoBWI] in {
def : Pat<(v4i32 (X86PAlignr VR128:$src1, VR128:$src2, (i8 imm:$imm))),
- (VPALIGNR128rr VR128:$src1, VR128:$src2, imm:$imm)>;
+ (VPALIGNRrri VR128:$src1, VR128:$src2, imm:$imm)>;
def : Pat<(v4f32 (X86PAlignr VR128:$src1, VR128:$src2, (i8 imm:$imm))),
- (VPALIGNR128rr VR128:$src1, VR128:$src2, imm:$imm)>;
+ (VPALIGNRrri VR128:$src1, VR128:$src2, imm:$imm)>;
def : Pat<(v8i16 (X86PAlignr VR128:$src1, VR128:$src2, (i8 imm:$imm))),
- (VPALIGNR128rr VR128:$src1, VR128:$src2, imm:$imm)>;
+ (VPALIGNRrri VR128:$src1, VR128:$src2, imm:$imm)>;
def : Pat<(v16i8 (X86PAlignr VR128:$src1, VR128:$src2, (i8 imm:$imm))),
- (VPALIGNR128rr VR128:$src1, VR128:$src2, imm:$imm)>;
+ (VPALIGNRrri VR128:$src1, VR128:$src2, imm:$imm)>;
}
let Predicates = [UseSSSE3] in {
def : Pat<(v4i32 (X86PAlignr VR128:$src1, VR128:$src2, (i8 imm:$imm))),
- (PALIGNR128rr VR128:$src1, VR128:$src2, imm:$imm)>;
+ (PALIGNRrri VR128:$src1, VR128:$src2, imm:$imm)>;
def : Pat<(v4f32 (X86PAlignr VR128:$src1, VR128:$src2, (i8 imm:$imm))),
- (PALIGNR128rr VR128:$src1, VR128:$src2, imm:$imm)>;
+ (PALIGNRrri VR128:$src1, VR128:$src2, imm:$imm)>;
def : Pat<(v8i16 (X86PAlignr VR128:$src1, VR128:$src2, (i8 imm:$imm))),
- (PALIGNR128rr VR128:$src1, VR128:$src2, imm:$imm)>;
+ (PALIGNRrri VR128:$src1, VR128:$src2, imm:$imm)>;
def : Pat<(v16i8 (X86PAlignr VR128:$src1, VR128:$src2, (i8 imm:$imm))),
- (PALIGNR128rr VR128:$src1, VR128:$src2, imm:$imm)>;
+ (PALIGNRrri VR128:$src1, VR128:$src2, imm:$imm)>;
}
//===---------------------------------------------------------------------===//
diff --git a/llvm/test/CodeGen/X86/sad.ll b/llvm/test/CodeGen/X86/sad.ll
index 89c3b527ae6..4e764a7d8e7 100644
--- a/llvm/test/CodeGen/X86/sad.ll
+++ b/llvm/test/CodeGen/X86/sad.ll
@@ -105,7 +105,7 @@ define i32 @sad_16i8() nounwind {
; AVX512BW-NEXT: vpaddd %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: vpunpckhqdq {{.*#+}} zmm1 = zmm0[1,1,3,3,5,5,7,7]
; AVX512BW-NEXT: vpaddd %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT: vpalignr $4, %zmm0, %zmm0, %zmm1
+; AVX512BW-NEXT: vpalignr {{.*#+}} zmm1 = zmm0[4,5,6,7,8,9,10,11,12,13,14,15,0,1,2,3,20,21,22,23,24,25,26,27,28,29,30,31,16,17,18,19,36,37,38,39,40,41,42,43,44,45,46,47,32,33,34,35,52,53,54,55,56,57,58,59,60,61,62,63,48,49,50,51]
; AVX512BW-NEXT: vpaddd %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: vmovd %xmm0, %eax
; AVX512BW-NEXT: retq
@@ -356,7 +356,7 @@ define i32 @sad_32i8() nounwind {
; AVX512BW-NEXT: vpaddd %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: vpunpckhqdq {{.*#+}} zmm1 = zmm0[1,1,3,3,5,5,7,7]
; AVX512BW-NEXT: vpaddd %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT: vpalignr $4, %zmm0, %zmm0, %zmm1
+; AVX512BW-NEXT: vpalignr {{.*#+}} zmm1 = zmm0[4,5,6,7,8,9,10,11,12,13,14,15,0,1,2,3,20,21,22,23,24,25,26,27,28,29,30,31,16,17,18,19,36,37,38,39,40,41,42,43,44,45,46,47,32,33,34,35,52,53,54,55,56,57,58,59,60,61,62,63,48,49,50,51]
; AVX512BW-NEXT: vpaddd %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: vmovd %xmm0, %eax
; AVX512BW-NEXT: retq
@@ -841,7 +841,7 @@ define i32 @sad_avx64i8() nounwind {
; AVX512BW-NEXT: vpaddd %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: vpunpckhqdq {{.*#+}} zmm1 = zmm0[1,1,3,3,5,5,7,7]
; AVX512BW-NEXT: vpaddd %zmm1, %zmm0, %zmm0
-; AVX512BW-NEXT: vpalignr $4, %zmm0, %zmm0, %zmm1
+; AVX512BW-NEXT: vpalignr {{.*#+}} zmm1 = zmm0[4,5,6,7,8,9,10,11,12,13,14,15,0,1,2,3,20,21,22,23,24,25,26,27,28,29,30,31,16,17,18,19,36,37,38,39,40,41,42,43,44,45,46,47,32,33,34,35,52,53,54,55,56,57,58,59,60,61,62,63,48,49,50,51]
; AVX512BW-NEXT: vpaddd %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: vmovd %xmm0, %eax
; AVX512BW-NEXT: retq
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll b/llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll
index 90892df18b3..9991bab046b 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll
@@ -13,7 +13,7 @@ define <64 x i8> @shuffle_v64i8_02_03_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_
;
; AVX512BW-LABEL: shuffle_v64i8_02_03_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u:
; AVX512BW: # BB#0:
-; AVX512BW-NEXT: vpalignr $2, %zmm0, %zmm0, %zmm0
+; AVX512BW-NEXT: vpalignr {{.*#+}} zmm0 = zmm0[2,3,4,5,6,7,8,9,10,11,12,13,14,15,0,1,18,19,20,21,22,23,24,25,26,27,28,29,30,31,16,17,34,35,36,37,38,39,40,41,42,43,44,45,46,47,32,33,50,51,52,53,54,55,56,57,58,59,60,61,62,63,48,49]
; AVX512BW-NEXT: retq
;
; AVX512DQ-LABEL: shuffle_v64i8_02_03_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u_u:
@@ -33,7 +33,7 @@ define <64 x i8> @shuffle_v64i8_79_00_01_02_03_04_05_06_07_08_09_10_11_12_13_14_
;
; AVX512BW-LABEL: shuffle_v64i8_79_00_01_02_03_04_05_06_07_08_09_10_11_12_13_14_95_16_17_18_19_20_21_22_23_24_25_26_27_28_29_30_111_32_33_34_35_36_37_38_39_40_41_42_43_44_45_46_127_48_49_50_51_52_53_54_55_56_57_58_59_60_61_62:
; AVX512BW: # BB#0:
-; AVX512BW-NEXT: vpalignr $15, %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT: vpalignr {{.*#+}} zmm0 = zmm1[15],zmm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14],zmm1[31],zmm0[16,17,18,19,20,21,22,23,24,25,26,27,28,29,30],zmm1[47],zmm0[32,33,34,35,36,37,38,39,40,41,42,43,44,45,46],zmm1[63],zmm0[48,49,50,51,52,53,54,55,56,57,58,59,60,61,62]
; AVX512BW-NEXT: retq
;
; AVX512DQ-LABEL: shuffle_v64i8_79_00_01_02_03_04_05_06_07_08_09_10_11_12_13_14_95_16_17_18_19_20_21_22_23_24_25_26_27_28_29_30_111_32_33_34_35_36_37_38_39_40_41_42_43_44_45_46_127_48_49_50_51_52_53_54_55_56_57_58_59_60_61_62:
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-v1.ll b/llvm/test/CodeGen/X86/vector-shuffle-v1.ll
index d99d80e97b9..22f5a0fba1d 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-v1.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-v1.ll
@@ -41,7 +41,7 @@ define <2 x i1> @shuf2i1_1_2(<2 x i1> %a) {
; VL_BW_DQ-NEXT: movb $1, %al
; VL_BW_DQ-NEXT: kmovb %eax, %k0
; VL_BW_DQ-NEXT: vpmovm2q %k0, %xmm1
-; VL_BW_DQ-NEXT: vpalignr $8, %xmm0, %xmm1, %xmm0
+; VL_BW_DQ-NEXT: vpalignr {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7]
; VL_BW_DQ-NEXT: vpsllq $63, %xmm0, %xmm0
; VL_BW_DQ-NEXT: vptestmq %xmm0, %xmm0, %k0
; VL_BW_DQ-NEXT: vpmovm2q %k0, %xmm0
OpenPOWER on IntegriCloud