XCHG (R32, EAX) - Latency

Experiment 1

• Instruction: XCHG R8D, EAX
• Chain instructions: MOVSX R8, R8D;MOVSX R8, R8D;MOVSX R8, R8D;MOVSX R8, R8D;MOVSX R8, R8D
• Chain latency: 5
• Code:

     0:	41 90                	xchg   r8d,eax
     2:	4d 63 c0             	movsxd r8,r8d
     5:	4d 63 c0             	movsxd r8,r8d
     8:	4d 63 c0             	movsxd r8,r8d
     b:	4d 63 c0             	movsxd r8,r8d
     e:	4d 63 c0             	movsxd r8,r8d
  

• Show nanoBench command
• Results:
  • Instructions retired: 6.0
  • Core cycles: 6.0
  • Reference cycles: 5.32
  • UOPS_EXECUTED.THREAD: 8.0

Experiment 2

• Instruction: XCHG R8D, EAX
• Code:

     0:	41 90                	xchg   r8d,eax
  

• Show nanoBench command
• Results:
  • Instructions retired: 1.0
  • Core cycles: 1.5
  • Reference cycles: 1.33
  • UOPS_EXECUTED.THREAD: 3.0

Experiment 3 (with dependency-breaking instructions)

• Instruction: XCHG R8D, EAX
• Chain instruction: MOV EAX, 0
• Code:

     0:	41 90                	xchg   r8d,eax
     2:	b8 00 00 00 00       	mov    eax,0x0
  

• Show nanoBench command
• Results:
  • Instructions retired: 2.0
  • Core cycles: 1.0
  • Reference cycles: 0.44
  • UOPS_EXECUTED.THREAD: 4.0

Experiment 1

• Instruction: XCHG R8D, EAX
• Chain instructions: MOVSX R8, EAX;MOVSX R8, R8D;MOVSX R8, R8D;MOVSX R8, R8D;MOVSX R8, R8D
• Chain latency: 5
• Code:

     0:	41 90                	xchg   r8d,eax
     2:	4c 63 c0             	movsxd r8,eax
     5:	4d 63 c0             	movsxd r8,r8d
     8:	4d 63 c0             	movsxd r8,r8d
     b:	4d 63 c0             	movsxd r8,r8d
     e:	4d 63 c0             	movsxd r8,r8d
  

• Show nanoBench command
• Results:
  • Instructions retired: 6.0
  • Core cycles: 7.0
  • Reference cycles: 5.99
  • UOPS_EXECUTED.THREAD: 8.0

Experiment 1

• Instruction: XCHG R8D, EAX
• Chain instructions: MOVSX RAX, R8D;MOVSX RAX, EAX;MOVSX RAX, EAX;MOVSX RAX, EAX;MOVSX RAX, EAX
• Chain latency: 5
• Code:

     0:	41 90                	xchg   r8d,eax
     2:	49 63 c0             	movsxd rax,r8d
     5:	48 63 c0             	movsxd rax,eax
     8:	48 63 c0             	movsxd rax,eax
     b:	48 63 c0             	movsxd rax,eax
     e:	48 63 c0             	movsxd rax,eax
  

• Show nanoBench command
• Results:
  • Instructions retired: 6.0
  • Core cycles: 6.0
  • Reference cycles: 5.32
  • UOPS_EXECUTED.THREAD: 8.0

Experiment 1

• Instruction: XCHG R8D, EAX
• Chain instructions: MOVSX RAX, EAX;MOVSX RAX, EAX;MOVSX RAX, EAX;MOVSX RAX, EAX;MOVSX RAX, EAX
• Chain latency: 5
• Code:

     0:	41 90                	xchg   r8d,eax
     2:	48 63 c0             	movsxd rax,eax
     5:	48 63 c0             	movsxd rax,eax
     8:	48 63 c0             	movsxd rax,eax
     b:	48 63 c0             	movsxd rax,eax
     e:	48 63 c0             	movsxd rax,eax
  

• Show nanoBench command
• Results:
  • Instructions retired: 6.0
  • Core cycles: 6.0
  • Reference cycles: 5.32
  • UOPS_EXECUTED.THREAD: 8.0

Experiment 2

• Instruction: XCHG R8D, EAX
• Code:

     0:	41 90                	xchg   r8d,eax
  

• Show nanoBench command
• Results:
  • Instructions retired: 1.0
  • Core cycles: 1.5
  • Reference cycles: 1.33
  • UOPS_EXECUTED.THREAD: 3.0

Experiment 3 (with dependency-breaking instructions)

• Instruction: XCHG R8D, EAX
• Chain instruction: MOV R8D, 0
• Code:

     0:	41 90                	xchg   r8d,eax
     2:	41 b8 00 00 00 00    	mov    r8d,0x0
  

• Show nanoBench command
• Results:
  • Instructions retired: 2.0
  • Core cycles: 1.0
  • Reference cycles: 0.67
  • UOPS_EXECUTED.THREAD: 4.0