[X86] Add FeatureCMOV to athlon and athlon-tbird cpus.

[craig.topper]

Summary

This matches gcc and one cpuid dump I found online. Given that these are considered 7th generation x86 CPU it seems likely they support cmov since cmov was added by Intel in their 6th generation.

[RKSimon]

LGTM - thanks

test/CodeGen/X86/select.ll

4 ↗

(On Diff #162571)

You might be able reduce the checks with more use of CHECK, CMOV and NO-CMOV common prefixes?

llvm/trunk/lib/Target/X86/X86.td

 
 // AMD CPUs.
 
 def : Proc<"k6",              [FeatureX87, FeatureSlowUAMem16, FeatureMMX]>;
 def : Proc<"k6-2",            [FeatureX87, FeatureSlowUAMem16, Feature3DNow]>;
 def : Proc<"k6-3",            [FeatureX87, FeatureSlowUAMem16, Feature3DNow]>;
 
 foreach P = ["athlon", "athlon-tbird"] in {
-  def : Proc<P, [FeatureX87, FeatureSlowUAMem16, Feature3DNowA,
+  def : Proc<P, [FeatureX87, FeatureSlowUAMem16, FeatureCMOV, Feature3DNowA,
                  FeatureNOPL, FeatureSlowSHLD]>;
 }
 
 foreach P = ["athlon-4", "athlon-xp", "athlon-mp"] in {
   def : Proc<P, [FeatureX87, FeatureSlowUAMem16, FeatureSSE1,
                  Feature3DNowA, FeatureFXSR, FeatureNOPL, FeatureSlowSHLD]>;
 }
 

llvm/trunk/test/CodeGen/X86/select.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-apple-darwin10 -mcpu=generic | FileCheck %s --check-prefix=CHECK --check-prefix=GENERIC
 ; RUN: llc < %s -mtriple=x86_64-apple-darwin10 -mcpu=atom    | FileCheck %s --check-prefix=CHECK --check-prefix=ATOM
+; RUN: llc < %s -mtriple=x86_64-apple-darwin10 -mcpu=athlon  | FileCheck %s --check-prefix=ATHLON
 ; RUN: llc < %s -mtriple=i386-intel-elfiamcu                 | FileCheck %s --check-prefix=MCU
 
 ; PR5757
 %0 = type { i64, i32 }
 
 define i32 @test1(%0* %p, %0* %q, i1 %r) nounwind {
 ; CHECK-LABEL: test1:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    addq $8, %rdi
 ; CHECK-NEXT:    addq $8, %rsi
 ; CHECK-NEXT:    testb $1, %dl
 ; CHECK-NEXT:    cmovneq %rdi, %rsi
 ; CHECK-NEXT:    movl (%rsi), %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: test1:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    addq $8, %rdi
+; ATHLON-NEXT:    addq $8, %rsi
+; ATHLON-NEXT:    testb $1, %dl
+; ATHLON-NEXT:    cmovneq %rdi, %rsi
+; ATHLON-NEXT:    movl (%rsi), %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test1:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    testb $1, %cl
 ; MCU-NEXT:    jne .LBB0_1
 ; MCU-NEXT:  # %bb.2:
 ; MCU-NEXT:    addl $8, %edx
 ; MCU-NEXT:    movl (%edx), %eax
 ; MCU-NEXT:    retl

 ; ATOM-NEXT:    jge LBB1_1
 ; ATOM-NEXT:  ## %bb.2: ## %bb91
 ; ATOM-NEXT:    xorl %eax, %eax
 ; ATOM-NEXT:    popq %rcx
 ; ATOM-NEXT:    retq
 ; ATOM-NEXT:  LBB1_1: ## %bb90
 ; ATOM-NEXT:    ud2
 ;
+; ATHLON-LABEL: test2:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    pushq %rax
+; ATHLON-NEXT:    callq _return_false
+; ATHLON-NEXT:    xorl %ecx, %ecx
+; ATHLON-NEXT:    testb $1, %al
+; ATHLON-NEXT:    movl $-480, %eax ## imm = 0xFE20
+; ATHLON-NEXT:    cmovnel %ecx, %eax
+; ATHLON-NEXT:    shll $3, %eax
+; ATHLON-NEXT:    cmpl $32768, %eax ## imm = 0x8000
+; ATHLON-NEXT:    jge LBB1_1
+; ATHLON-NEXT:  ## %bb.2: ## %bb91
+; ATHLON-NEXT:    xorl %eax, %eax
+; ATHLON-NEXT:    popq %rcx
+; ATHLON-NEXT:    retq
+; ATHLON-NEXT:  LBB1_1: ## %bb90
+; ATHLON-NEXT:    ud2
+;
 ; MCU-LABEL: test2:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    calll return_false
 ; MCU-NEXT:    xorl %ecx, %ecx
 ; MCU-NEXT:    testb $1, %al
 ; MCU-NEXT:    jne .LBB1_2
 ; MCU-NEXT:  # %bb.1: # %entry
 ; MCU-NEXT:    movl $-480, %ecx # imm = 0xFE20

 ; ATOM:       ## %bb.0: ## %entry
 ; ATOM-NEXT:    xorl %eax, %eax
 ; ATOM-NEXT:    leaq {{.*}}(%rip), %rcx
 ; ATOM-NEXT:    testl %edi, %edi
 ; ATOM-NEXT:    sete %al
 ; ATOM-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test3:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    xorl %eax, %eax
+; ATHLON-NEXT:    testl %edi, %edi
+; ATHLON-NEXT:    sete %al
+; ATHLON-NEXT:    leaq {{.*}}(%rip), %rcx
+; ATHLON-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test3:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    xorl %ecx, %ecx
 ; MCU-NEXT:    testl %eax, %eax
 ; MCU-NEXT:    sete %cl
 ; MCU-NEXT:    flds {{\.LCPI.*}}(,%ecx,4)
 ; MCU-NEXT:    retl
 entry:
   %0 = icmp eq i32 %x, 0
   %iftmp.0.0 = select i1 %0, float 4.200000e+01, float 2.300000e+01
   ret float %iftmp.0.0
 }
 
 define signext i8 @test4(i8* nocapture %P, double %F) nounwind readonly {
 ; CHECK-LABEL: test4:
 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    movsd {{.*#+}} xmm1 = mem[0],zero
 ; CHECK-NEXT:    xorl %eax, %eax
 ; CHECK-NEXT:    ucomisd %xmm0, %xmm1
 ; CHECK-NEXT:    seta %al
 ; CHECK-NEXT:    movsbl (%rdi,%rax,4), %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: test4:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    movsd {{.*#+}} xmm1 = mem[0],zero
+; ATHLON-NEXT:    xorl %eax, %eax
+; ATHLON-NEXT:    ucomisd %xmm0, %xmm1
+; ATHLON-NEXT:    seta %al
+; ATHLON-NEXT:    movsbl (%rdi,%rax,4), %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test4:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    movl %eax, %ecx
 ; MCU-NEXT:    fldl {{[0-9]+}}(%esp)
 ; MCU-NEXT:    flds {{\.LCPI.*}}
 ; MCU-NEXT:    fucompp
 ; MCU-NEXT:    fnstsw %ax
 ; MCU-NEXT:    xorl %edx, %edx

 ; CHECK-NEXT:  ## %bb.1:
 ; CHECK-NEXT:    movdqa %xmm1, %xmm0
 ; CHECK-NEXT:  LBB4_2:
 ; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
 ; CHECK-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
 ; CHECK-NEXT:    movd %xmm0, (%rsi)
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: test5:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    testb $1, %dil
+; ATHLON-NEXT:    jne LBB4_2
+; ATHLON-NEXT:  ## %bb.1:
+; ATHLON-NEXT:    movdqa %xmm1, %xmm0
+; ATHLON-NEXT:  LBB4_2:
+; ATHLON-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; ATHLON-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; ATHLON-NEXT:    movd %xmm0, (%rsi)
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test5:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    pushl %esi
 ; MCU-NEXT:    movl {{[0-9]+}}(%esp), %esi
 ; MCU-NEXT:    testb $1, %al
 ; MCU-NEXT:    jne .LBB4_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    movzwl {{[0-9]+}}(%esp), %ecx

 ; CHECK-NEXT:    movaps %xmm0, (%rsi)
 ; CHECK-NEXT:    retq
 ; CHECK-NEXT:  LBB5_1:
 ; CHECK-NEXT:    movaps (%rdx), %xmm0
 ; CHECK-NEXT:    mulps %xmm0, %xmm0
 ; CHECK-NEXT:    movaps %xmm0, (%rsi)
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: test6:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    testl %edi, %edi
+; ATHLON-NEXT:    je LBB5_1
+; ATHLON-NEXT:  ## %bb.2:
+; ATHLON-NEXT:    movaps (%rsi), %xmm0
+; ATHLON-NEXT:    movaps %xmm0, (%rsi)
+; ATHLON-NEXT:    retq
+; ATHLON-NEXT:  LBB5_1:
+; ATHLON-NEXT:    movaps (%rdx), %xmm0
+; ATHLON-NEXT:    mulps %xmm0, %xmm0
+; ATHLON-NEXT:    movaps %xmm0, (%rsi)
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test6:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    pushl %eax
 ; MCU-NEXT:    flds 12(%edx)
 ; MCU-NEXT:    fstps (%esp) # 4-byte Folded Spill
 ; MCU-NEXT:    flds 8(%edx)
 ; MCU-NEXT:    flds 4(%edx)
 ; MCU-NEXT:    flds (%ecx)

 ; ATOM-NEXT:    xorl %eax, %eax
 ; ATOM-NEXT:    leaq {{.*}}(%rip), %rcx
 ; ATOM-NEXT:    testl %edi, %edi
 ; ATOM-NEXT:    setns %al
 ; ATOM-NEXT:    shlq $4, %rax
 ; ATOM-NEXT:    fldt (%rax,%rcx)
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test7:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    xorl %eax, %eax
+; ATHLON-NEXT:    testl %edi, %edi
+; ATHLON-NEXT:    setns %al
+; ATHLON-NEXT:    shlq $4, %rax
+; ATHLON-NEXT:    leaq {{.*}}(%rip), %rcx
+; ATHLON-NEXT:    fldt (%rax,%rcx)
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test7:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    notl %eax
 ; MCU-NEXT:    shrl $27, %eax
 ; MCU-NEXT:    andl $-16, %eax
 ; MCU-NEXT:    fldt {{\.LCPI.*}}(%eax)
 ; MCU-NEXT:    retl
   %tmp9 = icmp sgt i32 %tmp8, -1

 ; ATOM-NEXT:    pcmpeqd %xmm2, %xmm2
 ; ATOM-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
 ; ATOM-NEXT:    paddd %xmm2, %xmm0
 ; ATOM-NEXT:    paddd %xmm2, %xmm1
 ; ATOM-NEXT:    movq %xmm1, 16(%rsi)
 ; ATOM-NEXT:    movdqa %xmm0, (%rsi)
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test8:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    testb $1, %dil
+; ATHLON-NEXT:    jne LBB7_1
+; ATHLON-NEXT:  ## %bb.2:
+; ATHLON-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; ATHLON-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; ATHLON-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; ATHLON-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; ATHLON-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; ATHLON-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; ATHLON-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; ATHLON-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; ATHLON-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; ATHLON-NEXT:    jmp LBB7_3
+; ATHLON-NEXT:  LBB7_1:
+; ATHLON-NEXT:    movd %r9d, %xmm0
+; ATHLON-NEXT:    movd %r8d, %xmm1
+; ATHLON-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; ATHLON-NEXT:    movd %ecx, %xmm2
+; ATHLON-NEXT:    movd %edx, %xmm0
+; ATHLON-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; ATHLON-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; ATHLON-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; ATHLON-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; ATHLON-NEXT:  LBB7_3:
+; ATHLON-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
+; ATHLON-NEXT:    pcmpeqd %xmm2, %xmm2
+; ATHLON-NEXT:    paddd %xmm2, %xmm0
+; ATHLON-NEXT:    paddd %xmm2, %xmm1
+; ATHLON-NEXT:    movq %xmm1, 16(%rsi)
+; ATHLON-NEXT:    movdqa %xmm0, (%rsi)
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test8:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    pushl %ebp
 ; MCU-NEXT:    pushl %ebx
 ; MCU-NEXT:    pushl %edi
 ; MCU-NEXT:    pushl %esi
 ; MCU-NEXT:    testb $1, %al
 ; MCU-NEXT:    jne .LBB7_1

 ; ATOM:       ## %bb.0:
 ; ATOM-NEXT:    cmpq $1, %rdi
 ; ATOM-NEXT:    sbbq %rax, %rax
 ; ATOM-NEXT:    orq %rsi, %rax
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test9:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    cmpq $1, %rdi
+; ATHLON-NEXT:    sbbq %rax, %rax
+; ATHLON-NEXT:    orq %rsi, %rax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test9:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    orl %edx, %eax
 ; MCU-NEXT:    jne .LBB8_1
 ; MCU-NEXT:  # %bb.2:
 ; MCU-NEXT:    movl $-1, %eax
 ; MCU-NEXT:    movl $-1, %edx
 ; MCU-NEXT:    retl

 ; ATOM:       ## %bb.0:
 ; ATOM-NEXT:    cmpq $1, %rdi
 ; ATOM-NEXT:    sbbq %rax, %rax
 ; ATOM-NEXT:    orq %rsi, %rax
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test9a:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    cmpq $1, %rdi
+; ATHLON-NEXT:    sbbq %rax, %rax
+; ATHLON-NEXT:    orq %rsi, %rax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test9a:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    orl %edx, %eax
 ; MCU-NEXT:    movl $-1, %eax
 ; MCU-NEXT:    movl $-1, %edx
 ; MCU-NEXT:    je .LBB9_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    movl {{[0-9]+}}(%esp), %eax

 ; ATOM:       ## %bb.0:
 ; ATOM-NEXT:    cmpq $1, %rdi
 ; ATOM-NEXT:    sbbq %rax, %rax
 ; ATOM-NEXT:    orq %rsi, %rax
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test9b:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    cmpq $1, %rdi
+; ATHLON-NEXT:    sbbq %rax, %rax
+; ATHLON-NEXT:    orq %rsi, %rax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test9b:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    movl %edx, %ecx
 ; MCU-NEXT:    xorl %edx, %edx
 ; MCU-NEXT:    orl %ecx, %eax
 ; MCU-NEXT:    sete %dl
 ; MCU-NEXT:    negl %edx
 ; MCU-NEXT:    movl {{[0-9]+}}(%esp), %eax

 ; CHECK-LABEL: test10:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    xorl %eax, %eax
 ; CHECK-NEXT:    testq %rdi, %rdi
 ; CHECK-NEXT:    setne %al
 ; CHECK-NEXT:    leaq -1(%rax,%rax), %rax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: test10:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    xorl %eax, %eax
+; ATHLON-NEXT:    testq %rdi, %rdi
+; ATHLON-NEXT:    setne %al
+; ATHLON-NEXT:    leaq -1(%rax,%rax), %rax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test10:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    orl %edx, %eax
 ; MCU-NEXT:    movl $-1, %eax
 ; MCU-NEXT:    movl $-1, %edx
 ; MCU-NEXT:    je .LBB11_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    xorl %edx, %edx

 ; CHECK-LABEL: test11:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    cmpq $1, %rdi
 ; CHECK-NEXT:    sbbq %rax, %rax
 ; CHECK-NEXT:    notq %rax
 ; CHECK-NEXT:    orq %rsi, %rax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: test11:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    cmpq $1, %rdi
+; ATHLON-NEXT:    sbbq %rax, %rax
+; ATHLON-NEXT:    notq %rax
+; ATHLON-NEXT:    orq %rsi, %rax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test11:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    orl %edx, %eax
 ; MCU-NEXT:    je .LBB12_1
 ; MCU-NEXT:  # %bb.2:
 ; MCU-NEXT:    movl $-1, %eax
 ; MCU-NEXT:    movl $-1, %edx
 ; MCU-NEXT:    retl

 ; CHECK-LABEL: test11a:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    cmpq $1, %rdi
 ; CHECK-NEXT:    sbbq %rax, %rax
 ; CHECK-NEXT:    notq %rax
 ; CHECK-NEXT:    orq %rsi, %rax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: test11a:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    cmpq $1, %rdi
+; ATHLON-NEXT:    sbbq %rax, %rax
+; ATHLON-NEXT:    notq %rax
+; ATHLON-NEXT:    orq %rsi, %rax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test11a:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    orl %edx, %eax
 ; MCU-NEXT:    movl $-1, %eax
 ; MCU-NEXT:    movl $-1, %edx
 ; MCU-NEXT:    jne .LBB13_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    movl {{[0-9]+}}(%esp), %eax

 ; ATOM-NEXT:    cmpl %esi, %edi
 ; ATOM-NEXT:    sbbl %eax, %eax
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test13:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    cmpl %esi, %edi
+; ATHLON-NEXT:    sbbl %eax, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test13:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    cmpl %edx, %eax
 ; MCU-NEXT:    sbbl %eax, %eax
 ; MCU-NEXT:    retl
   %c = icmp ult i32 %a, %b
   %d = sext i1 %c to i32
   ret i32 %d
 }
 
 define i32 @test14(i32 %a, i32 %b) nounwind {
 ; CHECK-LABEL: test14:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    xorl %eax, %eax
 ; CHECK-NEXT:    cmpl %esi, %edi
 ; CHECK-NEXT:    setae %al
 ; CHECK-NEXT:    negl %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: test14:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    xorl %eax, %eax
+; ATHLON-NEXT:    cmpl %esi, %edi
+; ATHLON-NEXT:    setae %al
+; ATHLON-NEXT:    negl %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test14:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    xorl %ecx, %ecx
 ; MCU-NEXT:    cmpl %edx, %eax
 ; MCU-NEXT:    setae %cl
 ; MCU-NEXT:    negl %ecx
 ; MCU-NEXT:    movl %ecx, %eax
 ; MCU-NEXT:    retl

 ; ATOM-NEXT:    negl %edi
 ; ATOM-NEXT:    sbbl %eax, %eax
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test15:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    negl %edi
+; ATHLON-NEXT:    sbbl %eax, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test15:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    negl %eax
 ; MCU-NEXT:    sbbl %eax, %eax
 ; MCU-NEXT:    retl
 entry:
   %cmp = icmp ne i32 %x, 0
   %sub = sext i1 %cmp to i32

 ; ATOM-NEXT:    negq %rdi
 ; ATOM-NEXT:    sbbq %rax, %rax
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test16:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    negq %rdi
+; ATHLON-NEXT:    sbbq %rax, %rax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test16:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    movl %eax, %ecx
 ; MCU-NEXT:    xorl %eax, %eax
 ; MCU-NEXT:    orl %edx, %ecx
 ; MCU-NEXT:    setne %al
 ; MCU-NEXT:    negl %eax
 ; MCU-NEXT:    movl %eax, %edx

 ; ATOM-NEXT:    sbbl %eax, %eax
 ; ATOM-NEXT:    ## kill: def $ax killed $ax killed $eax
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test17:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    negw %di
+; ATHLON-NEXT:    sbbl %eax, %eax
+; ATHLON-NEXT:    ## kill: def $ax killed $ax killed $eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test17:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    negw %ax
 ; MCU-NEXT:    sbbl %eax, %eax
 ; MCU-NEXT:    # kill: def $ax killed $ax killed $eax
 ; MCU-NEXT:    retl
 entry:
   %cmp = icmp ne i16 %x, 0

 ; ATOM:       ## %bb.0:
 ; ATOM-NEXT:    cmpl $15, %edi
 ; ATOM-NEXT:    cmovgel %edx, %esi
 ; ATOM-NEXT:    movl %esi, %eax
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    nop
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: test18:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    cmpl $15, %edi
+; ATHLON-NEXT:    cmovgel %edx, %esi
+; ATHLON-NEXT:    movl %esi, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test18:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    cmpl $15, %eax
 ; MCU-NEXT:    jl .LBB19_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    movl %ecx, %edx
 ; MCU-NEXT:  .LBB19_2:
 ; MCU-NEXT:    movl %edx, %eax
 ; MCU-NEXT:    retl
   %cmp = icmp slt i32 %x, 15
   %sel = select i1 %cmp, i8 %a, i8 %b
   ret i8 %sel
 }
 
 define i32 @trunc_select_miscompile(i32 %a, i1 zeroext %cc) {
 ; CHECK-LABEL: trunc_select_miscompile:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    orb $2, %sil
 ; CHECK-NEXT:    movl %esi, %ecx
 ; CHECK-NEXT:    shll %cl, %edi
 ; CHECK-NEXT:    movl %edi, %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: trunc_select_miscompile:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    orb $2, %sil
+; ATHLON-NEXT:    movl %esi, %ecx
+; ATHLON-NEXT:    shll %cl, %edi
+; ATHLON-NEXT:    movl %edi, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: trunc_select_miscompile:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    orb $2, %dl
 ; MCU-NEXT:    movl %edx, %ecx
 ; MCU-NEXT:    shll %cl, %eax
 ; MCU-NEXT:    retl
   %tmp1 = select i1 %cc, i32 3, i32 2
   %tmp2 = shl i32 %a, %tmp1

 ; ATOM-NEXT:    cmpl $-128, %eax
 ; ATOM-NEXT:    jl LBB21_2
 ; ATOM-NEXT:  ## %bb.1:
 ; ATOM-NEXT:    movl %eax, %ecx
 ; ATOM-NEXT:  LBB21_2:
 ; ATOM-NEXT:    movb %cl, (%rsi)
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: clamp_i8:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    cmpl $127, %edi
+; ATHLON-NEXT:    movl $127, %eax
+; ATHLON-NEXT:    cmovlel %edi, %eax
+; ATHLON-NEXT:    cmpl $-128, %eax
+; ATHLON-NEXT:    movb $-128, %cl
+; ATHLON-NEXT:    jl LBB21_2
+; ATHLON-NEXT:  ## %bb.1:
+; ATHLON-NEXT:    movl %eax, %ecx
+; ATHLON-NEXT:  LBB21_2:
+; ATHLON-NEXT:    movb %cl, (%rsi)
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: clamp_i8:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    cmpl $127, %eax
 ; MCU-NEXT:    movl $127, %ecx
 ; MCU-NEXT:    jg .LBB21_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    movl %eax, %ecx
 ; MCU-NEXT:  .LBB21_2:

 ; ATOM-NEXT:    movl $32767, %eax ## imm = 0x7FFF
 ; ATOM-NEXT:    movl $32768, %ecx ## imm = 0x8000
 ; ATOM-NEXT:    cmovlel %edi, %eax
 ; ATOM-NEXT:    cmpl $-32768, %eax ## imm = 0x8000
 ; ATOM-NEXT:    cmovgel %eax, %ecx
 ; ATOM-NEXT:    movw %cx, (%rsi)
 ; ATOM-NEXT:    retq
 ;
+; ATHLON-LABEL: clamp:
+; ATHLON:       ## %bb.0:
+; ATHLON-NEXT:    cmpl $32767, %edi ## imm = 0x7FFF
+; ATHLON-NEXT:    movl $32767, %eax ## imm = 0x7FFF
+; ATHLON-NEXT:    cmovlel %edi, %eax
+; ATHLON-NEXT:    cmpl $-32768, %eax ## imm = 0x8000
+; ATHLON-NEXT:    movl $32768, %ecx ## imm = 0x8000
+; ATHLON-NEXT:    cmovgel %eax, %ecx
+; ATHLON-NEXT:    movw %cx, (%rsi)
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: clamp:
 ; MCU:       # %bb.0:
 ; MCU-NEXT:    cmpl $32767, %eax # imm = 0x7FFF
 ; MCU-NEXT:    movl $32767, %ecx # imm = 0x7FFF
 ; MCU-NEXT:    jg .LBB22_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    movl %eax, %ecx
 ; MCU-NEXT:  .LBB22_2:

 ; CHECK-NEXT:  LBB23_3: ## %CF242
 ; CHECK-NEXT:    ## =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    cmpl %eax, %eax
 ; CHECK-NEXT:    ucomiss %xmm0, %xmm0
 ; CHECK-NEXT:    jp LBB23_3
 ; CHECK-NEXT:  ## %bb.4: ## %CF244
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: test19:
+; ATHLON:       ## %bb.0: ## %BB
+; ATHLON-NEXT:    movl $-1, %eax
+; ATHLON-NEXT:    movb $1, %cl
+; ATHLON-NEXT:    .p2align 4, 0x90
+; ATHLON-NEXT:  LBB23_1: ## %CF
+; ATHLON-NEXT:    ## =>This Inner Loop Header: Depth=1
+; ATHLON-NEXT:    testb %cl, %cl
+; ATHLON-NEXT:    jne LBB23_1
+; ATHLON-NEXT:  ## %bb.2: ## %CF250
+; ATHLON-NEXT:    ## in Loop: Header=BB23_1 Depth=1
+; ATHLON-NEXT:    jne LBB23_1
+; ATHLON-NEXT:    .p2align 4, 0x90
+; ATHLON-NEXT:  LBB23_3: ## %CF242
+; ATHLON-NEXT:    ## =>This Inner Loop Header: Depth=1
+; ATHLON-NEXT:    cmpl %eax, %eax
+; ATHLON-NEXT:    ucomiss %xmm0, %xmm0
+; ATHLON-NEXT:    jp LBB23_3
+; ATHLON-NEXT:  ## %bb.4: ## %CF244
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: test19:
 ; MCU:       # %bb.0: # %BB
 ; MCU-NEXT:    movl $-1, %ecx
 ; MCU-NEXT:    movb $1, %al
 ; MCU-NEXT:    .p2align 4, 0x90
 ; MCU-NEXT:  .LBB23_1: # %CF
 ; MCU-NEXT:    # =>This Inner Loop Header: Depth=1
 ; MCU-NEXT:    testb %al, %al

 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    movl %edi, %eax
 ; CHECK-NEXT:    xorl $43, %eax
 ; CHECK-NEXT:    testb $1, %sil
 ; CHECK-NEXT:    cmovel %edi, %eax
 ; CHECK-NEXT:    ## kill: def $ax killed $ax killed $eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: select_xor_1:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    movl %edi, %eax
+; ATHLON-NEXT:    xorl $43, %eax
+; ATHLON-NEXT:    testb $1, %sil
+; ATHLON-NEXT:    cmovel %edi, %eax
+; ATHLON-NEXT:    ## kill: def $ax killed $ax killed $eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: select_xor_1:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    andl $1, %edx
 ; MCU-NEXT:    negl %edx
 ; MCU-NEXT:    andl $43, %edx
 ; MCU-NEXT:    xorl %edx, %eax
 ; MCU-NEXT:    # kill: def $ax killed $ax killed $eax
 ; MCU-NEXT:    retl

 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    movl %edi, %eax
 ; CHECK-NEXT:    xorl $43, %eax
 ; CHECK-NEXT:    testb $1, %sil
 ; CHECK-NEXT:    cmovel %edi, %eax
 ; CHECK-NEXT:    ## kill: def $ax killed $ax killed $eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: select_xor_1b:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    movl %edi, %eax
+; ATHLON-NEXT:    xorl $43, %eax
+; ATHLON-NEXT:    testb $1, %sil
+; ATHLON-NEXT:    cmovel %edi, %eax
+; ATHLON-NEXT:    ## kill: def $ax killed $ax killed $eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: select_xor_1b:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    testb $1, %dl
 ; MCU-NEXT:    je .LBB25_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    xorl $43, %eax
 ; MCU-NEXT:  .LBB25_2: # %entry
 ; MCU-NEXT:    # kill: def $ax killed $ax killed $eax

 ; CHECK-LABEL: select_xor_2:
 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    xorl %edi, %esi
 ; CHECK-NEXT:    testb $1, %dl
 ; CHECK-NEXT:    cmovel %edi, %esi
 ; CHECK-NEXT:    movl %esi, %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: select_xor_2:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    xorl %edi, %esi
+; ATHLON-NEXT:    testb $1, %dl
+; ATHLON-NEXT:    cmovel %edi, %esi
+; ATHLON-NEXT:    movl %esi, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: select_xor_2:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    andl $1, %ecx
 ; MCU-NEXT:    negl %ecx
 ; MCU-NEXT:    andl %edx, %ecx
 ; MCU-NEXT:    xorl %ecx, %eax
 ; MCU-NEXT:    retl
 entry:

 ; CHECK-LABEL: select_xor_2b:
 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    xorl %edi, %esi
 ; CHECK-NEXT:    testb $1, %dl
 ; CHECK-NEXT:    cmovel %edi, %esi
 ; CHECK-NEXT:    movl %esi, %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: select_xor_2b:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    xorl %edi, %esi
+; ATHLON-NEXT:    testb $1, %dl
+; ATHLON-NEXT:    cmovel %edi, %esi
+; ATHLON-NEXT:    movl %esi, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: select_xor_2b:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    testb $1, %cl
 ; MCU-NEXT:    je .LBB27_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    xorl %edx, %eax
 ; MCU-NEXT:  .LBB27_2: # %entry
 ; MCU-NEXT:    retl

 ; CHECK-LABEL: select_or:
 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    orl %edi, %esi
 ; CHECK-NEXT:    testb $1, %dl
 ; CHECK-NEXT:    cmovel %edi, %esi
 ; CHECK-NEXT:    movl %esi, %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: select_or:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    orl %edi, %esi
+; ATHLON-NEXT:    testb $1, %dl
+; ATHLON-NEXT:    cmovel %edi, %esi
+; ATHLON-NEXT:    movl %esi, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: select_or:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    andl $1, %ecx
 ; MCU-NEXT:    negl %ecx
 ; MCU-NEXT:    andl %edx, %ecx
 ; MCU-NEXT:    orl %ecx, %eax
 ; MCU-NEXT:    retl
 entry:

 ; CHECK-LABEL: select_or_b:
 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    orl %edi, %esi
 ; CHECK-NEXT:    testb $1, %dl
 ; CHECK-NEXT:    cmovel %edi, %esi
 ; CHECK-NEXT:    movl %esi, %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: select_or_b:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    orl %edi, %esi
+; ATHLON-NEXT:    testb $1, %dl
+; ATHLON-NEXT:    cmovel %edi, %esi
+; ATHLON-NEXT:    movl %esi, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: select_or_b:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    testb $1, %cl
 ; MCU-NEXT:    je .LBB29_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    orl %edx, %eax
 ; MCU-NEXT:  .LBB29_2: # %entry
 ; MCU-NEXT:    retl

 ; CHECK-LABEL: select_or_1:
 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    orl %edi, %esi
 ; CHECK-NEXT:    testb $1, %dl
 ; CHECK-NEXT:    cmovel %edi, %esi
 ; CHECK-NEXT:    movl %esi, %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: select_or_1:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    orl %edi, %esi
+; ATHLON-NEXT:    testb $1, %dl
+; ATHLON-NEXT:    cmovel %edi, %esi
+; ATHLON-NEXT:    movl %esi, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: select_or_1:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    andl $1, %ecx
 ; MCU-NEXT:    negl %ecx
 ; MCU-NEXT:    andl %edx, %ecx
 ; MCU-NEXT:    orl %ecx, %eax
 ; MCU-NEXT:    retl
 entry:

 ; CHECK-LABEL: select_or_1b:
 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    orl %edi, %esi
 ; CHECK-NEXT:    testb $1, %dl
 ; CHECK-NEXT:    cmovel %edi, %esi
 ; CHECK-NEXT:    movl %esi, %eax
 ; CHECK-NEXT:    retq
 ;
+; ATHLON-LABEL: select_or_1b:
+; ATHLON:       ## %bb.0: ## %entry
+; ATHLON-NEXT:    orl %edi, %esi
+; ATHLON-NEXT:    testb $1, %dl
+; ATHLON-NEXT:    cmovel %edi, %esi
+; ATHLON-NEXT:    movl %esi, %eax
+; ATHLON-NEXT:    retq
+;
 ; MCU-LABEL: select_or_1b:
 ; MCU:       # %bb.0: # %entry
 ; MCU-NEXT:    testb $1, %cl
 ; MCU-NEXT:    je .LBB31_2
 ; MCU-NEXT:  # %bb.1:
 ; MCU-NEXT:    orl %edx, %eax
 ; MCU-NEXT:  .LBB31_2: # %entry
 ; MCU-NEXT:    retl
 entry:
  %and = and i32 %cond, 1
  %cmp10 = icmp ne i32 %and, 1
  %0 = or i32 %B, %A
  %1 = select i1 %cmp10, i32 %A, i32 %0
  ret i32 %1
 }