shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity

[X86] Use a two lea sequence for multiply by 37, 41, and 73.

Browse Source 
These fit a pattern used by 11, 21, and 19.

llvm-svn: 337871
This commit is contained in:
Craig Topper 2018-07-24 23:44:17 +00:00
parent b5342b592e
commit 56c104f104
5 changed files with 114 additions and 59 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
llvm/lib/Target/X86/X86ISelLowering.cpp
View File

@ -33716,6 +33716,9 @@ static SDValue combineMulSpecial(uint64_t MulAmt, SDNode *N, SelectionDAG &DAG,
case 21:
// mul x, 21 => add ((shl (mul x, 5), 2), x)
return combineMulShlAddOrSub(5, 2, /*isAdd*/ true);
case 41:
// mul x, 41 => add ((shl (mul x, 5), 3), x)
return combineMulShlAddOrSub(5, 3, /*isAdd*/ true);
case 22:
// mul x, 22 => add (add ((shl (mul x, 5), 2), x), x)
return DAG.getNode(ISD::ADD, DL, VT, N->getOperand(0),
@ -33723,6 +33726,12 @@ static SDValue combineMulSpecial(uint64_t MulAmt, SDNode *N, SelectionDAG &DAG,
case 19:
// mul x, 19 => add ((shl (mul x, 9), 1), x)
return combineMulShlAddOrSub(9, 1, /*isAdd*/ true);
case 37:
// mul x, 37 => add ((shl (mul x, 9), 2), x)
return combineMulShlAddOrSub(9, 2, /*isAdd*/ true);
case 73:
// mul x, 73 => add ((shl (mul x, 9), 3), x)
return combineMulShlAddOrSub(9, 3, /*isAdd*/ true);
case 13:
// mul x, 13 => add ((shl (mul x, 3), 2), x)
return combineMulShlAddOrSub(3, 2, /*isAdd*/ true);

74
llvm/test/CodeGen/X86/funnel-shift.ll
View File

@ -127,31 +127,33 @@ define i37 @fshl_i37(i37 %x, i37 %y, i37 %z) nounwind {
; X64-AVX2-LABEL: fshl_i37:
; X64-AVX2: # %bb.0:
; X64-AVX2-NEXT: pushq %rbx
; X64-AVX2-NEXT: movq %rdx, %r9
; X64-AVX2-NEXT: movabsq $137438953471, %r10 # imm = 0x1FFFFFFFFF
; X64-AVX2-NEXT: andq %r10, %rsi
; X64-AVX2-NEXT: movl $37, %r8d
; X64-AVX2-NEXT: subq %rdx, %r8
; X64-AVX2-NEXT: andq %r10, %r9
; X64-AVX2-NEXT: movq %rdx, %r10
; X64-AVX2-NEXT: movabsq $137438953471, %r8 # imm = 0x1FFFFFFFFF
; X64-AVX2-NEXT: andq %r8, %rsi
; X64-AVX2-NEXT: movl $37, %r9d
; X64-AVX2-NEXT: subq %rdx, %r9
; X64-AVX2-NEXT: andq %r8, %r10
; X64-AVX2-NEXT: movabsq $-2492803253203993461, %r11 # imm = 0xDD67C8A60DD67C8B
; X64-AVX2-NEXT: movq %r9, %rax
; X64-AVX2-NEXT: movq %r10, %rax
; X64-AVX2-NEXT: mulq %r11
; X64-AVX2-NEXT: shrq $5, %rdx
; X64-AVX2-NEXT: imulq $37, %rdx, %rax
; X64-AVX2-NEXT: subq %rax, %r9
; X64-AVX2-NEXT: leaq (%rdx,%rdx,8), %rax
; X64-AVX2-NEXT: leaq (%rdx,%rax,4), %rax
; X64-AVX2-NEXT: subq %rax, %r10
; X64-AVX2-NEXT: movq %rdi, %rbx
; X64-AVX2-NEXT: movl %r9d, %ecx
; X64-AVX2-NEXT: movl %r10d, %ecx
; X64-AVX2-NEXT: shlq %cl, %rbx
; X64-AVX2-NEXT: andq %r9, %r8
; X64-AVX2-NEXT: movq %r8, %rax
; X64-AVX2-NEXT: andq %r10, %rax
; X64-AVX2-NEXT: mulq %r11
; X64-AVX2-NEXT: shrl $5, %edx
; X64-AVX2-NEXT: imull $37, %edx, %eax
; X64-AVX2-NEXT: subl %eax, %r8d
; X64-AVX2-NEXT: movl %r8d, %ecx
; X64-AVX2-NEXT: shrq $5, %rdx
; X64-AVX2-NEXT: leaq (%rdx,%rdx,8), %rax
; X64-AVX2-NEXT: leal (%rdx,%rax,4), %eax
; X64-AVX2-NEXT: subl %eax, %r9d
; X64-AVX2-NEXT: movl %r9d, %ecx
; X64-AVX2-NEXT: shrq %cl, %rsi
; X64-AVX2-NEXT: orq %rbx, %rsi
; X64-AVX2-NEXT: testq %r9, %r9
; X64-AVX2-NEXT: testq %r10, %r10
; X64-AVX2-NEXT: cmoveq %rdi, %rsi
; X64-AVX2-NEXT: movq %rsi, %rax
; X64-AVX2-NEXT: popq %rbx
@ -359,31 +361,33 @@ define i37 @fshr_i37(i37 %x, i37 %y, i37 %z) nounwind {
; X64-AVX2-LABEL: fshr_i37:
; X64-AVX2: # %bb.0:
; X64-AVX2-NEXT: pushq %rbx
; X64-AVX2-NEXT: movq %rdx, %r9
; X64-AVX2-NEXT: movabsq $137438953471, %r11 # imm = 0x1FFFFFFFFF
; X64-AVX2-NEXT: movq %rsi, %r10
; X64-AVX2-NEXT: andq %r11, %r10
; X64-AVX2-NEXT: movl $37, %r8d
; X64-AVX2-NEXT: subq %rdx, %r8
; X64-AVX2-NEXT: andq %r11, %r9
; X64-AVX2-NEXT: movq %rdx, %r10
; X64-AVX2-NEXT: movabsq $137438953471, %r8 # imm = 0x1FFFFFFFFF
; X64-AVX2-NEXT: movq %rsi, %r11
; X64-AVX2-NEXT: andq %r8, %r11
; X64-AVX2-NEXT: movl $37, %r9d
; X64-AVX2-NEXT: subq %rdx, %r9
; X64-AVX2-NEXT: andq %r8, %r10
; X64-AVX2-NEXT: movabsq $-2492803253203993461, %rbx # imm = 0xDD67C8A60DD67C8B
; X64-AVX2-NEXT: movq %r9, %rax
; X64-AVX2-NEXT: movq %r10, %rax
; X64-AVX2-NEXT: mulq %rbx
; X64-AVX2-NEXT: shrq $5, %rdx
; X64-AVX2-NEXT: imulq $37, %rdx, %rax
; X64-AVX2-NEXT: subq %rax, %r9
; X64-AVX2-NEXT: movl %r9d, %ecx
; X64-AVX2-NEXT: shrq %cl, %r10
; X64-AVX2-NEXT: leaq (%rdx,%rdx,8), %rax
; X64-AVX2-NEXT: leaq (%rdx,%rax,4), %rax
; X64-AVX2-NEXT: subq %rax, %r10
; X64-AVX2-NEXT: movl %r10d, %ecx
; X64-AVX2-NEXT: shrq %cl, %r11
; X64-AVX2-NEXT: andq %r9, %r8
; X64-AVX2-NEXT: movq %r8, %rax
; X64-AVX2-NEXT: andq %r11, %rax
; X64-AVX2-NEXT: mulq %rbx
; X64-AVX2-NEXT: shrl $5, %edx
; X64-AVX2-NEXT: imull $37, %edx, %eax
; X64-AVX2-NEXT: subl %eax, %r8d
; X64-AVX2-NEXT: movl %r8d, %ecx
; X64-AVX2-NEXT: shrq $5, %rdx
; X64-AVX2-NEXT: leaq (%rdx,%rdx,8), %rax
; X64-AVX2-NEXT: leal (%rdx,%rax,4), %eax
; X64-AVX2-NEXT: subl %eax, %r9d
; X64-AVX2-NEXT: movl %r9d, %ecx
; X64-AVX2-NEXT: shlq %cl, %rdi
; X64-AVX2-NEXT: orq %r10, %rdi
; X64-AVX2-NEXT: testq %r9, %r9
; X64-AVX2-NEXT: orq %r11, %rdi
; X64-AVX2-NEXT: testq %r10, %r10
; X64-AVX2-NEXT: cmoveq %rsi, %rdi
; X64-AVX2-NEXT: movq %rdi, %rax
; X64-AVX2-NEXT: popq %rbx

24
llvm/test/CodeGen/X86/mul-constant-i16.ll
View File

@ -642,13 +642,17 @@ define i16 @test_mul_by_32(i16 %x) {
define i16 @test_mul_by_37(i16 %x) {
; X86-LABEL: test_mul_by_37:
; X86: # %bb.0:
; X86-NEXT: imull $37, {{[0-9]+}}(%esp), %eax
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: leal (%eax,%eax,8), %ecx
; X86-NEXT: leal (%eax,%ecx,4), %eax
; X86-NEXT: # kill: def $ax killed $ax killed $eax
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_37:
; X64: # %bb.0:
; X64-NEXT: imull $37, %edi, %eax
; X64-NEXT: # kill: def $edi killed $edi def $rdi
; X64-NEXT: leal (%rdi,%rdi,8), %eax
; X64-NEXT: leal (%rdi,%rax,4), %eax
; X64-NEXT: # kill: def $ax killed $ax killed $eax
; X64-NEXT: retq
%mul = mul nsw i16 %x, 37
@ -658,13 +662,17 @@ define i16 @test_mul_by_37(i16 %x) {
define i16 @test_mul_by_41(i16 %x) {
; X86-LABEL: test_mul_by_41:
; X86: # %bb.0:
; X86-NEXT: imull $41, {{[0-9]+}}(%esp), %eax
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: leal (%eax,%eax,4), %ecx
; X86-NEXT: leal (%eax,%ecx,8), %eax
; X86-NEXT: # kill: def $ax killed $ax killed $eax
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_41:
; X64: # %bb.0:
; X64-NEXT: imull $41, %edi, %eax
; X64-NEXT: # kill: def $edi killed $edi def $rdi
; X64-NEXT: leal (%rdi,%rdi,4), %eax
; X64-NEXT: leal (%rdi,%rax,8), %eax
; X64-NEXT: # kill: def $ax killed $ax killed $eax
; X64-NEXT: retq
%mul = mul nsw i16 %x, 41
@ -697,13 +705,17 @@ define i16 @test_mul_by_62(i16 %x) {
define i16 @test_mul_by_73(i16 %x) {
; X86-LABEL: test_mul_by_73:
; X86: # %bb.0:
; X86-NEXT: imull $73, {{[0-9]+}}(%esp), %eax
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: leal (%eax,%eax,8), %ecx
; X86-NEXT: leal (%eax,%ecx,8), %eax
; X86-NEXT: # kill: def $ax killed $ax killed $eax
; X86-NEXT: retl
;
; X64-LABEL: test_mul_by_73:
; X64: # %bb.0:
; X64-NEXT: imull $73, %edi, %eax
; X64-NEXT: # kill: def $edi killed $edi def $rdi
; X64-NEXT: leal (%rdi,%rdi,8), %eax
; X64-NEXT: leal (%rdi,%rax,8), %eax
; X64-NEXT: # kill: def $ax killed $ax killed $eax
; X64-NEXT: retq
%mul = mul nsw i16 %x, 73

36
llvm/test/CodeGen/X86/mul-constant-i32.ll
View File

@ -1679,17 +1679,23 @@ define i32 @test_mul_by_32(i32 %x) {
define i32 @test_mul_by_37(i32 %x) {
; X86-LABEL: test_mul_by_37:
; X86: # %bb.0:
; X86-NEXT: imull $37, {{[0-9]+}}(%esp), %eax
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: leal (%eax,%eax,8), %ecx
; X86-NEXT: leal (%eax,%ecx,4), %eax
; X86-NEXT: retl
;
; X64-HSW-LABEL: test_mul_by_37:
; X64-HSW: # %bb.0:
; X64-HSW-NEXT: imull $37, %edi, %eax # sched: [3:1.00]
; X64-HSW-NEXT: # kill: def $edi killed $edi def $rdi
; X64-HSW-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
; X64-HSW-NEXT: leal (%rdi,%rax,4), %eax # sched: [1:0.50]
; X64-HSW-NEXT: retq # sched: [7:1.00]
;
; X64-JAG-LABEL: test_mul_by_37:
; X64-JAG: # %bb.0:
; X64-JAG-NEXT: imull $37, %edi, %eax # sched: [3:1.00]
; X64-JAG-NEXT: # kill: def $edi killed $edi def $rdi
; X64-JAG-NEXT: leal (%rdi,%rdi,8), %eax # sched: [2:1.00]
; X64-JAG-NEXT: leal (%rdi,%rax,4), %eax # sched: [2:1.00]
; X64-JAG-NEXT: retq # sched: [4:1.00]
;
; X86-NOOPT-LABEL: test_mul_by_37:
@ -1723,17 +1729,23 @@ define i32 @test_mul_by_37(i32 %x) {
define i32 @test_mul_by_41(i32 %x) {
; X86-LABEL: test_mul_by_41:
; X86: # %bb.0:
; X86-NEXT: imull $41, {{[0-9]+}}(%esp), %eax
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: leal (%eax,%eax,4), %ecx
; X86-NEXT: leal (%eax,%ecx,8), %eax
; X86-NEXT: retl
;
; X64-HSW-LABEL: test_mul_by_41:
; X64-HSW: # %bb.0:
; X64-HSW-NEXT: imull $41, %edi, %eax # sched: [3:1.00]
; X64-HSW-NEXT: # kill: def $edi killed $edi def $rdi
; X64-HSW-NEXT: leal (%rdi,%rdi,4), %eax # sched: [1:0.50]
; X64-HSW-NEXT: leal (%rdi,%rax,8), %eax # sched: [1:0.50]
; X64-HSW-NEXT: retq # sched: [7:1.00]
;
; X64-JAG-LABEL: test_mul_by_41:
; X64-JAG: # %bb.0:
; X64-JAG-NEXT: imull $41, %edi, %eax # sched: [3:1.00]
; X64-JAG-NEXT: # kill: def $edi killed $edi def $rdi
; X64-JAG-NEXT: leal (%rdi,%rdi,4), %eax # sched: [2:1.00]
; X64-JAG-NEXT: leal (%rdi,%rax,8), %eax # sched: [2:1.00]
; X64-JAG-NEXT: retq # sched: [4:1.00]
;
; X86-NOOPT-LABEL: test_mul_by_41:
@ -1824,17 +1836,23 @@ define i32 @test_mul_by_62(i32 %x) {
define i32 @test_mul_by_73(i32 %x) {
; X86-LABEL: test_mul_by_73:
; X86: # %bb.0:
; X86-NEXT: imull $73, {{[0-9]+}}(%esp), %eax
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: leal (%eax,%eax,8), %ecx
; X86-NEXT: leal (%eax,%ecx,8), %eax
; X86-NEXT: retl
;
; X64-HSW-LABEL: test_mul_by_73:
; X64-HSW: # %bb.0:
; X64-HSW-NEXT: imull $73, %edi, %eax # sched: [3:1.00]
; X64-HSW-NEXT: # kill: def $edi killed $edi def $rdi
; X64-HSW-NEXT: leal (%rdi,%rdi,8), %eax # sched: [1:0.50]
; X64-HSW-NEXT: leal (%rdi,%rax,8), %eax # sched: [1:0.50]
; X64-HSW-NEXT: retq # sched: [7:1.00]
;
; X64-JAG-LABEL: test_mul_by_73:
; X64-JAG: # %bb.0:
; X64-JAG-NEXT: imull $73, %edi, %eax # sched: [3:1.00]
; X64-JAG-NEXT: # kill: def $edi killed $edi def $rdi
; X64-JAG-NEXT: leal (%rdi,%rdi,8), %eax # sched: [2:1.00]
; X64-JAG-NEXT: leal (%rdi,%rax,8), %eax # sched: [2:1.00]
; X64-JAG-NEXT: retq # sched: [4:1.00]
;
; X86-NOOPT-LABEL: test_mul_by_73:

30
llvm/test/CodeGen/X86/mul-constant-i64.ll
View File

@ -1764,20 +1764,24 @@ define i64 @test_mul_by_32(i64 %x) {
define i64 @test_mul_by_37(i64 %x) {
; X86-LABEL: test_mul_by_37:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: leal (%eax,%eax,8), %ecx
; X86-NEXT: leal (%eax,%ecx,4), %ecx
; X86-NEXT: movl $37, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
; X86-NEXT: imull $37, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
; X64-HSW-LABEL: test_mul_by_37:
; X64-HSW: # %bb.0:
; X64-HSW-NEXT: imulq $37, %rdi, %rax # sched: [3:1.00]
; X64-HSW-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
; X64-HSW-NEXT: leaq (%rdi,%rax,4), %rax # sched: [1:0.50]
; X64-HSW-NEXT: retq # sched: [7:1.00]
;
; X64-JAG-LABEL: test_mul_by_37:
; X64-JAG: # %bb.0:
; X64-JAG-NEXT: imulq $37, %rdi, %rax # sched: [6:4.00]
; X64-JAG-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [2:1.00]
; X64-JAG-NEXT: leaq (%rdi,%rax,4), %rax # sched: [2:1.00]
; X64-JAG-NEXT: retq # sched: [4:1.00]
;
; X86-NOOPT-LABEL: test_mul_by_37:
@ -1814,20 +1818,24 @@ define i64 @test_mul_by_37(i64 %x) {
define i64 @test_mul_by_41(i64 %x) {
; X86-LABEL: test_mul_by_41:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: leal (%eax,%eax,4), %ecx
; X86-NEXT: leal (%eax,%ecx,8), %ecx
; X86-NEXT: movl $41, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
; X86-NEXT: imull $41, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
; X64-HSW-LABEL: test_mul_by_41:
; X64-HSW: # %bb.0:
; X64-HSW-NEXT: imulq $41, %rdi, %rax # sched: [3:1.00]
; X64-HSW-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [1:0.50]
; X64-HSW-NEXT: leaq (%rdi,%rax,8), %rax # sched: [1:0.50]
; X64-HSW-NEXT: retq # sched: [7:1.00]
;
; X64-JAG-LABEL: test_mul_by_41:
; X64-JAG: # %bb.0:
; X64-JAG-NEXT: imulq $41, %rdi, %rax # sched: [6:4.00]
; X64-JAG-NEXT: leaq (%rdi,%rdi,4), %rax # sched: [2:1.00]
; X64-JAG-NEXT: leaq (%rdi,%rax,8), %rax # sched: [2:1.00]
; X64-JAG-NEXT: retq # sched: [4:1.00]
;
; X86-NOOPT-LABEL: test_mul_by_41:
@ -1927,20 +1935,24 @@ define i64 @test_mul_by_62(i64 %x) {
define i64 @test_mul_by_73(i64 %x) {
; X86-LABEL: test_mul_by_73:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: leal (%eax,%eax,8), %ecx
; X86-NEXT: leal (%eax,%ecx,8), %ecx
; X86-NEXT: movl $73, %eax
; X86-NEXT: mull {{[0-9]+}}(%esp)
; X86-NEXT: imull $73, {{[0-9]+}}(%esp), %ecx
; X86-NEXT: addl %ecx, %edx
; X86-NEXT: retl
;
; X64-HSW-LABEL: test_mul_by_73:
; X64-HSW: # %bb.0:
; X64-HSW-NEXT: imulq $73, %rdi, %rax # sched: [3:1.00]
; X64-HSW-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [1:0.50]
; X64-HSW-NEXT: leaq (%rdi,%rax,8), %rax # sched: [1:0.50]
; X64-HSW-NEXT: retq # sched: [7:1.00]
;
; X64-JAG-LABEL: test_mul_by_73:
; X64-JAG: # %bb.0:
; X64-JAG-NEXT: imulq $73, %rdi, %rax # sched: [6:4.00]
; X64-JAG-NEXT: leaq (%rdi,%rdi,8), %rax # sched: [2:1.00]
; X64-JAG-NEXT: leaq (%rdi,%rax,8), %rax # sched: [2:1.00]
; X64-JAG-NEXT: retq # sched: [4:1.00]
;
; X86-NOOPT-LABEL: test_mul_by_73: