VMOVDQA64 (YMM, K, M256) - Port Usage

With blocking instructions for port '23':

Code:

   0:	45 8b 46 40          	mov    r8d,DWORD PTR [r14+0x40]
   4:	45 8b 4e 44          	mov    r9d,DWORD PTR [r14+0x44]
   8:	45 8b 56 48          	mov    r10d,DWORD PTR [r14+0x48]
   c:	45 8b 5e 4c          	mov    r11d,DWORD PTR [r14+0x4c]
  10:	45 8b 66 50          	mov    r12d,DWORD PTR [r14+0x50]
  14:	41 8b 5e 54          	mov    ebx,DWORD PTR [r14+0x54]
  18:	41 8b 4e 58          	mov    ecx,DWORD PTR [r14+0x58]
  1c:	41 8b 56 5c          	mov    edx,DWORD PTR [r14+0x5c]
  20:	45 8b 46 40          	mov    r8d,DWORD PTR [r14+0x40]
  24:	45 8b 4e 44          	mov    r9d,DWORD PTR [r14+0x44]
  28:	62 d1 fd 29 6f 06    	vmovdqa64 ymm0{k1},YMMWORD PTR [r14]

Init:

MOV R15, 10000;
L: VADDPS YMM0, YMM1, YMM1;
VADDPS YMM0, YMM1, YMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS YMM0, YMM0, YMM0

Show nanoBench command
Results:
- Instructions retired: 11.0
- Core cycles: 5.62
- Reference cycles: 2.43
- UOPS_PORT_23: 11.0

⇨ One μop that can only use port '23'

With blocking instructions for ports {'1', '5'}:

Code:

   0:	47 8d 04 6e          	lea    r8d,[r14+r13*2]
   4:	47 8d 0c 6e          	lea    r9d,[r14+r13*2]
   8:	47 8d 14 6e          	lea    r10d,[r14+r13*2]
   c:	47 8d 1c 6e          	lea    r11d,[r14+r13*2]
  10:	47 8d 24 6e          	lea    r12d,[r14+r13*2]
  14:	43 8d 1c 6e          	lea    ebx,[r14+r13*2]
  18:	43 8d 0c 6e          	lea    ecx,[r14+r13*2]
  1c:	43 8d 14 6e          	lea    edx,[r14+r13*2]
  20:	47 8d 04 6e          	lea    r8d,[r14+r13*2]
  24:	47 8d 0c 6e          	lea    r9d,[r14+r13*2]
  28:	62 d1 fd 29 6f 06    	vmovdqa64 ymm0{k1},YMMWORD PTR [r14]

Init:

MOV R15, 10000;
L: VADDPS YMM0, YMM1, YMM1;
VADDPS YMM0, YMM1, YMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS YMM0, YMM0, YMM0

Show nanoBench command
Results:
- Instructions retired: 11.0
- Core cycles: 5.08
- Reference cycles: 2.43
- UOPS_PORT_5: 5.0
- UOPS_PORT_1: 5.02

With blocking instructions for ports {'0', '1', '5'}:

Code:

   0:	c5 e9 55 cb          	vandnpd xmm1,xmm2,xmm3
   4:	c5 e9 55 e3          	vandnpd xmm4,xmm2,xmm3
   8:	c5 e9 55 eb          	vandnpd xmm5,xmm2,xmm3
   c:	c5 e9 55 f3          	vandnpd xmm6,xmm2,xmm3
  10:	c5 e9 55 fb          	vandnpd xmm7,xmm2,xmm3
  14:	c5 69 55 c3          	vandnpd xmm8,xmm2,xmm3
  18:	c5 69 55 cb          	vandnpd xmm9,xmm2,xmm3
  1c:	c5 69 55 d3          	vandnpd xmm10,xmm2,xmm3
  20:	c5 69 55 db          	vandnpd xmm11,xmm2,xmm3
  24:	c5 69 55 e3          	vandnpd xmm12,xmm2,xmm3
  28:	62 d1 fd 29 6f 06    	vmovdqa64 ymm0{k1},YMMWORD PTR [r14]

Init:

MOV R15, 10000;
L: VADDPS YMM0, YMM1, YMM1;
VADDPS YMM0, YMM1, YMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS XMM2, XMM2, XMM2;
VXORPS XMM3, XMM3, XMM3;
VXORPS YMM0, YMM0, YMM0

Show nanoBench command
Results:
- Instructions retired: 11.0
- Core cycles: 3.79
- Reference cycles: 2.08
- UOPS_PORT_5: 3.66
- UOPS_PORT_0: 3.63
- UOPS_PORT_1: 3.7

⇨ One μop that can only use ports {'0', '1', '5'}

With an indexed addressing mode

With blocking instructions for port '23':

Code:

   0:	45 8b 46 40          	mov    r8d,DWORD PTR [r14+0x40]
   4:	45 8b 4e 44          	mov    r9d,DWORD PTR [r14+0x44]
   8:	45 8b 56 48          	mov    r10d,DWORD PTR [r14+0x48]
   c:	45 8b 5e 4c          	mov    r11d,DWORD PTR [r14+0x4c]
  10:	45 8b 66 50          	mov    r12d,DWORD PTR [r14+0x50]
  14:	41 8b 5e 54          	mov    ebx,DWORD PTR [r14+0x54]
  18:	41 8b 4e 58          	mov    ecx,DWORD PTR [r14+0x58]
  1c:	41 8b 56 5c          	mov    edx,DWORD PTR [r14+0x5c]
  20:	45 8b 46 40          	mov    r8d,DWORD PTR [r14+0x40]
  24:	45 8b 4e 44          	mov    r9d,DWORD PTR [r14+0x44]
  28:	62 91 fd 29 6f 04 2e 	vmovdqa64 ymm0{k1},YMMWORD PTR [r14+r13*1]

Init:

MOV R15, 10000;
L: VADDPS YMM0, YMM1, YMM1;
VADDPS YMM0, YMM1, YMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS YMM0, YMM0, YMM0;
XOR R13, R13

Show nanoBench command
Results:
- Instructions retired: 11.0
- Core cycles: 5.62
- Reference cycles: 2.43
- UOPS_PORT_23: 11.0

⇨ One μop that can only use port '23'

With blocking instructions for ports {'1', '5'}:

Code:

   0:	47 8d 04 6e          	lea    r8d,[r14+r13*2]
   4:	47 8d 0c 6e          	lea    r9d,[r14+r13*2]
   8:	47 8d 14 6e          	lea    r10d,[r14+r13*2]
   c:	47 8d 1c 6e          	lea    r11d,[r14+r13*2]
  10:	47 8d 24 6e          	lea    r12d,[r14+r13*2]
  14:	43 8d 1c 6e          	lea    ebx,[r14+r13*2]
  18:	43 8d 0c 6e          	lea    ecx,[r14+r13*2]
  1c:	43 8d 14 6e          	lea    edx,[r14+r13*2]
  20:	47 8d 04 6e          	lea    r8d,[r14+r13*2]
  24:	47 8d 0c 6e          	lea    r9d,[r14+r13*2]
  28:	62 91 fd 29 6f 04 2e 	vmovdqa64 ymm0{k1},YMMWORD PTR [r14+r13*1]

Init:

MOV R15, 10000;
L: VADDPS YMM0, YMM1, YMM1;
VADDPS YMM0, YMM1, YMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS YMM0, YMM0, YMM0;
XOR R13, R13

Show nanoBench command
Results:
- Instructions retired: 11.0
- Core cycles: 5.08
- Reference cycles: 2.43
- UOPS_PORT_5: 5.0
- UOPS_PORT_1: 5.02

With blocking instructions for ports {'0', '1', '5'}:

Code:

   0:	c5 e9 55 cb          	vandnpd xmm1,xmm2,xmm3
   4:	c5 e9 55 e3          	vandnpd xmm4,xmm2,xmm3
   8:	c5 e9 55 eb          	vandnpd xmm5,xmm2,xmm3
   c:	c5 e9 55 f3          	vandnpd xmm6,xmm2,xmm3
  10:	c5 e9 55 fb          	vandnpd xmm7,xmm2,xmm3
  14:	c5 69 55 c3          	vandnpd xmm8,xmm2,xmm3
  18:	c5 69 55 cb          	vandnpd xmm9,xmm2,xmm3
  1c:	c5 69 55 d3          	vandnpd xmm10,xmm2,xmm3
  20:	c5 69 55 db          	vandnpd xmm11,xmm2,xmm3
  24:	c5 69 55 e3          	vandnpd xmm12,xmm2,xmm3
  28:	62 91 fd 29 6f 04 2e 	vmovdqa64 ymm0{k1},YMMWORD PTR [r14+r13*1]

Init:

MOV R15, 10000;
L: VADDPS YMM0, YMM1, YMM1;
VADDPS YMM0, YMM1, YMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS XMM2, XMM2, XMM2;
VXORPS XMM3, XMM3, XMM3;
VXORPS YMM0, YMM0, YMM0;
XOR R13, R13

Show nanoBench command
Results:
- Instructions retired: 11.0
- Core cycles: 3.77
- Reference cycles: 1.73
- UOPS_PORT_5: 3.67
- UOPS_PORT_0: 3.65
- UOPS_PORT_1: 3.67

⇨ One μop that can only use ports {'0', '1', '5'}