MOVBE (M32, R32) - Latency

Experiment 1 (with R8D=0)

• Instruction: MOVBE dword ptr [R14], R8D
• Chain instructions: MOV R12D, [R14];MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;XOR R14, R12; XOR R14, R12
• Chain latency: ≥6
• Code:

     0:	45 0f 38 f1 06       	movbe  DWORD PTR [r14],r8d
     5:	45 8b 26             	mov    r12d,DWORD PTR [r14]
     8:	4d 63 e4             	movsxd r12,r12d
     b:	4d 63 e4             	movsxd r12,r12d
     e:	4d 63 e4             	movsxd r12,r12d
    11:	4d 63 e4             	movsxd r12,r12d
    14:	4d 31 e6             	xor    r14,r12
    17:	4d 31 e6             	xor    r14,r12
  

• Init:

  MOV R8D, 0

• Show nanoBench command
• Results:
  • Instructions retired: 8.0
  • Core cycles: 2.35
  • Reference cycles: 1.04
  • UOPS_EXECUTED.THREAD: 10.0

Experiment 2 (with R8D=1)

• Instruction: MOVBE dword ptr [R14], R8D
• Chain instructions: MOV R12D, [R14];MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;XOR R14, R12; XOR R14, R12
• Chain latency: ≥6
• Code:

     0:	45 0f 38 f1 06       	movbe  DWORD PTR [r14],r8d
     5:	45 8b 26             	mov    r12d,DWORD PTR [r14]
     8:	4d 63 e4             	movsxd r12,r12d
     b:	4d 63 e4             	movsxd r12,r12d
     e:	4d 63 e4             	movsxd r12,r12d
    11:	4d 63 e4             	movsxd r12,r12d
    14:	4d 31 e6             	xor    r14,r12
    17:	4d 31 e6             	xor    r14,r12
  

• Init:

  MOV R8D, 1

• Show nanoBench command
• Results:
  • Instructions retired: 8.0
  • Core cycles: 2.34
  • Reference cycles: 2.08
  • UOPS_EXECUTED.THREAD: 10.0

Experiment 3 (with R8D=2)

• Instruction: MOVBE dword ptr [R14], R8D
• Chain instructions: MOV R12D, [R14];MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;XOR R14, R12; XOR R14, R12
• Chain latency: ≥6
• Code:

     0:	45 0f 38 f1 06       	movbe  DWORD PTR [r14],r8d
     5:	45 8b 26             	mov    r12d,DWORD PTR [r14]
     8:	4d 63 e4             	movsxd r12,r12d
     b:	4d 63 e4             	movsxd r12,r12d
     e:	4d 63 e4             	movsxd r12,r12d
    11:	4d 63 e4             	movsxd r12,r12d
    14:	4d 31 e6             	xor    r14,r12
    17:	4d 31 e6             	xor    r14,r12
  

• Init:

  MOV R8D, 2

• Show nanoBench command
• Results:
  • Instructions retired: 8.0
  • Core cycles: 2.35
  • Reference cycles: 1.91
  • UOPS_EXECUTED.THREAD: 10.0

Experiment 4 (with additional nop, with R8D=0)

• Instruction: MOVBE dword ptr [R14], R8D
• Chain instructions: MOV R12D, [R14];MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;XOR R14, R12; XOR R14, R12;nop
• Chain latency: ≥6
• Code:

     0:	45 0f 38 f1 06       	movbe  DWORD PTR [r14],r8d
     5:	45 8b 26             	mov    r12d,DWORD PTR [r14]
     8:	4d 63 e4             	movsxd r12,r12d
     b:	4d 63 e4             	movsxd r12,r12d
     e:	4d 63 e4             	movsxd r12,r12d
    11:	4d 63 e4             	movsxd r12,r12d
    14:	4d 31 e6             	xor    r14,r12
    17:	4d 31 e6             	xor    r14,r12
    1a:	90                   	nop
  

• Init:

  MOV R8D, 0

• Show nanoBench command
• Results:
  • Instructions retired: 9.0
  • Core cycles: 3.15
  • Reference cycles: 1.56
  • UOPS_EXECUTED.THREAD: 10.0

Experiment 5 (with additional nop, with R8D=1)

• Instruction: MOVBE dword ptr [R14], R8D
• Chain instructions: MOV R12D, [R14];MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;XOR R14, R12; XOR R14, R12;nop
• Chain latency: ≥6
• Code:

     0:	45 0f 38 f1 06       	movbe  DWORD PTR [r14],r8d
     5:	45 8b 26             	mov    r12d,DWORD PTR [r14]
     8:	4d 63 e4             	movsxd r12,r12d
     b:	4d 63 e4             	movsxd r12,r12d
     e:	4d 63 e4             	movsxd r12,r12d
    11:	4d 63 e4             	movsxd r12,r12d
    14:	4d 31 e6             	xor    r14,r12
    17:	4d 31 e6             	xor    r14,r12
    1a:	90                   	nop
  

• Init:

  MOV R8D, 1

• Show nanoBench command
• Results:
  • Instructions retired: 9.0
  • Core cycles: 3.15
  • Reference cycles: 1.04
  • UOPS_EXECUTED.THREAD: 10.0

Experiment 6 (with additional nop, with R8D=2)

• Instruction: MOVBE dword ptr [R14], R8D
• Chain instructions: MOV R12D, [R14];MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;XOR R14, R12; XOR R14, R12;nop
• Chain latency: ≥6
• Code:

     0:	45 0f 38 f1 06       	movbe  DWORD PTR [r14],r8d
     5:	45 8b 26             	mov    r12d,DWORD PTR [r14]
     8:	4d 63 e4             	movsxd r12,r12d
     b:	4d 63 e4             	movsxd r12,r12d
     e:	4d 63 e4             	movsxd r12,r12d
    11:	4d 63 e4             	movsxd r12,r12d
    14:	4d 31 e6             	xor    r14,r12
    17:	4d 31 e6             	xor    r14,r12
    1a:	90                   	nop
  

• Init:

  MOV R8D, 2

• Show nanoBench command
• Results:
  • Instructions retired: 9.0
  • Core cycles: 3.16
  • Reference cycles: 2.6
  • UOPS_EXECUTED.THREAD: 10.0

Experiment 1

• Instruction: MOVBE dword ptr [R14+R13], R8D
• Chain instructions: MOV R12D, [R14+R13];MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;XOR R13, R12; XOR R13, R12
• Chain latency: ≥6
• Code:

     0:	47 0f 38 f1 04 2e    	movbe  DWORD PTR [r14+r13*1],r8d
     6:	47 8b 24 2e          	mov    r12d,DWORD PTR [r14+r13*1]
     a:	4d 63 e4             	movsxd r12,r12d
     d:	4d 63 e4             	movsxd r12,r12d
    10:	4d 63 e4             	movsxd r12,r12d
    13:	4d 63 e4             	movsxd r12,r12d
    16:	4d 31 e5             	xor    r13,r12
    19:	4d 31 e5             	xor    r13,r12
  

• Init:

  XOR R13, R13

• Show nanoBench command
• Results:
  • Instructions retired: 8.0
  • Core cycles: 17.05
  • Reference cycles: 8.32
  • UOPS_EXECUTED.THREAD: 10.0

Experiment 2 (with additional nop)

• Instruction: MOVBE dword ptr [R14+R13], R8D
• Chain instructions: MOV R12D, [R14+R13];MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;MOVSX R12, R12D;XOR R13, R12; XOR R13, R12;nop
• Chain latency: ≥6
• Code:

     0:	47 0f 38 f1 04 2e    	movbe  DWORD PTR [r14+r13*1],r8d
     6:	47 8b 24 2e          	mov    r12d,DWORD PTR [r14+r13*1]
     a:	4d 63 e4             	movsxd r12,r12d
     d:	4d 63 e4             	movsxd r12,r12d
    10:	4d 63 e4             	movsxd r12,r12d
    13:	4d 63 e4             	movsxd r12,r12d
    16:	4d 31 e5             	xor    r13,r12
    19:	4d 31 e5             	xor    r13,r12
    1c:	90                   	nop
  

• Init:

  XOR R13, R13

• Show nanoBench command
• Results:
  • Instructions retired: 9.0
  • Core cycles: 17.06
  • Reference cycles: 8.32
  • UOPS_EXECUTED.THREAD: 10.0

Experiment 1

• Instruction: MOVBE dword ptr [R14], R8D
• Chain instructions: MOV R8D, [R14];MOVSX R8, R8D;MOVSX R8, R8D;MOVSX R8, R8D;MOVSX R8, R8D
• Chain latency: ≥4
• Code:

     0:	45 0f 38 f1 06       	movbe  DWORD PTR [r14],r8d
     5:	45 8b 06             	mov    r8d,DWORD PTR [r14]
     8:	4d 63 c0             	movsxd r8,r8d
     b:	4d 63 c0             	movsxd r8,r8d
     e:	4d 63 c0             	movsxd r8,r8d
    11:	4d 63 c0             	movsxd r8,r8d
  

• Show nanoBench command
• Results:
  • Instructions retired: 6.0
  • Core cycles: 5.0
  • Reference cycles: 2.6
  • UOPS_EXECUTED.THREAD: 8.0