VPMULLQ (YMM, YMM, YMM) - Port Usage

With blocking instructions for port '0':

Code:

   0:	41 0f 7e c0          	movd   r8d,mm0
   4:	41 0f 7e c1          	movd   r9d,mm0
   8:	41 0f 7e c2          	movd   r10d,mm0
   c:	41 0f 7e c3          	movd   r11d,mm0
  10:	41 0f 7e c4          	movd   r12d,mm0
  14:	0f 7e c3             	movd   ebx,mm0
  17:	0f 7e c1             	movd   ecx,mm0
  1a:	0f 7e c2             	movd   edx,mm0
  1d:	41 0f 7e c0          	movd   r8d,mm0
  21:	41 0f 7e c1          	movd   r9d,mm0
  25:	62 f2 f5 28 40 c2    	vpmullq ymm0,ymm1,ymm2

Init:

MOV R15, 10000;
L: VADDPS YMM0, YMM1, YMM1;
VADDPS YMM0, YMM1, YMM1;
DEC R15;
JNZ L;
PXOR MM0, MM0;
VZEROALL;
VXORPS YMM1, YMM1, YMM1;
VXORPS YMM2, YMM2, YMM2

Show nanoBench command
Results:
- Instructions retired: 11.0
- Core cycles: 10.08
- Reference cycles: 5.84
- UOPS_PORT_0: 9.99

With blocking instructions for port '1':

Code:

   0:	44 8d 05 80 00 00 00 	lea    r8d,[rip+0x80]        # 0x87
   7:	44 8d 0d 80 00 00 00 	lea    r9d,[rip+0x80]        # 0x8e
   e:	44 8d 15 80 00 00 00 	lea    r10d,[rip+0x80]        # 0x95
  15:	44 8d 1d 80 00 00 00 	lea    r11d,[rip+0x80]        # 0x9c
  1c:	44 8d 25 80 00 00 00 	lea    r12d,[rip+0x80]        # 0xa3
  23:	8d 1d 80 00 00 00    	lea    ebx,[rip+0x80]        # 0xa9
  29:	8d 0d 80 00 00 00    	lea    ecx,[rip+0x80]        # 0xaf
  2f:	8d 15 80 00 00 00    	lea    edx,[rip+0x80]        # 0xb5
  35:	44 8d 05 80 00 00 00 	lea    r8d,[rip+0x80]        # 0xbc
  3c:	44 8d 0d 80 00 00 00 	lea    r9d,[rip+0x80]        # 0xc3
  43:	62 f2 f5 28 40 c2    	vpmullq ymm0,ymm1,ymm2

Init:

MOV R15, 10000;
L: VADDPS YMM0, YMM1, YMM1;
VADDPS YMM0, YMM1, YMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS YMM1, YMM1, YMM1;
VXORPS YMM2, YMM2, YMM2

Show nanoBench command
Results:
- Instructions retired: 11.0
- Core cycles: 10.04
- Reference cycles: 5.84
- UOPS_PORT_1: 10.0

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

Code:

   0:	c5 d9 58 dd          	vaddpd xmm3,xmm4,xmm5
   4:	c5 d9 58 f5          	vaddpd xmm6,xmm4,xmm5
   8:	c5 d9 58 fd          	vaddpd xmm7,xmm4,xmm5
   c:	c5 59 58 c5          	vaddpd xmm8,xmm4,xmm5
  10:	c5 59 58 cd          	vaddpd xmm9,xmm4,xmm5
  14:	c5 59 58 d5          	vaddpd xmm10,xmm4,xmm5
  18:	c5 59 58 dd          	vaddpd xmm11,xmm4,xmm5
  1c:	c5 59 58 e5          	vaddpd xmm12,xmm4,xmm5
  20:	c5 d9 58 dd          	vaddpd xmm3,xmm4,xmm5
  24:	c5 d9 58 f5          	vaddpd xmm6,xmm4,xmm5
  28:	62 f2 f5 28 40 c2    	vpmullq ymm0,ymm1,ymm2

Init:

MOV R15, 10000;
L: VADDPS YMM0, YMM1, YMM1;
VADDPS YMM0, YMM1, YMM1;
DEC R15;
JNZ L;
VZEROALL;
MOV RAX, 0x4000000040000000;
MOV [R14+0], RAX;
MOV [R14+8], RAX;
VMOVUPD XMM4, [R14];
VMOVUPD XMM5, [R14];
VXORPS YMM1, YMM1, YMM1;
VXORPS YMM2, YMM2, YMM2

Show nanoBench command
Results:
- Instructions retired: 11.0
- Core cycles: 6.68
- Reference cycles: 3.77
- UOPS_PORT_1: 6.5
- UOPS_PORT_0: 6.49

⇨ 3 μops that can only use ports {'0', '1'}