VDPBF16PS (ZMM, ZMM, ZMM) - 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:	0f 7e c0             	movd   eax,mm0
  20:	41 0f 7e c0          	movd   r8d,mm0
  24:	41 0f 7e c1          	movd   r9d,mm0
  28:	41 0f 7e c2          	movd   r10d,mm0
  2c:	41 0f 7e c3          	movd   r11d,mm0
  30:	41 0f 7e c4          	movd   r12d,mm0
  34:	0f 7e c3             	movd   ebx,mm0
  37:	0f 7e c1             	movd   ecx,mm0
  3a:	62 f2 76 48 52 c2    	vdpbf16ps zmm0,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
PXOR MM0, MM0;
VZEROALL;
MOV RAX, 0x4000000040000000;
MOV [R14+0], RAX;
MOV [R14+8], RAX;
MOV [R14+16], RAX;
MOV [R14+24], RAX;
MOV [R14+32], RAX;
MOV [R14+40], RAX;
MOV [R14+48], RAX;
MOV [R14+56], RAX;
VMOVUPD ZMM0, [R14];
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 17.0
- Core cycles: 16.02
- Reference cycles: 10.32
- UOPS_PORT_0: 16.0

With blocking instructions for port '5':

Code:

   0:	41 0f 6e c0          	movd   mm0,r8d
   4:	41 0f 6e c8          	movd   mm1,r8d
   8:	41 0f 6e d0          	movd   mm2,r8d
   c:	41 0f 6e d8          	movd   mm3,r8d
  10:	41 0f 6e e0          	movd   mm4,r8d
  14:	41 0f 6e e8          	movd   mm5,r8d
  18:	41 0f 6e f0          	movd   mm6,r8d
  1c:	41 0f 6e f8          	movd   mm7,r8d
  20:	41 0f 6e c0          	movd   mm0,r8d
  24:	41 0f 6e c8          	movd   mm1,r8d
  28:	41 0f 6e d0          	movd   mm2,r8d
  2c:	41 0f 6e d8          	movd   mm3,r8d
  30:	41 0f 6e e0          	movd   mm4,r8d
  34:	41 0f 6e e8          	movd   mm5,r8d
  38:	41 0f 6e f0          	movd   mm6,r8d
  3c:	41 0f 6e f8          	movd   mm7,r8d
  40:	62 f2 76 48 52 c2    	vdpbf16ps zmm0,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
MOV RAX, 0x4000000040000000;
MOV [R14+0], RAX;
MOV [R14+8], RAX;
MOV [R14+16], RAX;
MOV [R14+24], RAX;
MOV [R14+32], RAX;
MOV [R14+40], RAX;
MOV [R14+48], RAX;
MOV [R14+56], RAX;
VMOVUPD ZMM0, [R14];
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 17.0
- Core cycles: 18.08
- Reference cycles: 11.61
- UOPS_PORT_5B: 18.0
- UOPS_PORT_5B>=2: 0.0

⇨ 2 μops that can only use port '5'

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

Code:

   0:	c4 e2 59 de dd       	vaesdec xmm3,xmm4,xmm5
   5:	c4 e2 59 de f5       	vaesdec xmm6,xmm4,xmm5
   a:	c4 e2 59 de fd       	vaesdec xmm7,xmm4,xmm5
   f:	c4 62 59 de c5       	vaesdec xmm8,xmm4,xmm5
  14:	c4 62 59 de cd       	vaesdec xmm9,xmm4,xmm5
  19:	c4 62 59 de d5       	vaesdec xmm10,xmm4,xmm5
  1e:	c4 62 59 de dd       	vaesdec xmm11,xmm4,xmm5
  23:	c4 62 59 de e5       	vaesdec xmm12,xmm4,xmm5
  28:	c4 e2 59 de dd       	vaesdec xmm3,xmm4,xmm5
  2d:	c4 e2 59 de f5       	vaesdec xmm6,xmm4,xmm5
  32:	c4 e2 59 de fd       	vaesdec xmm7,xmm4,xmm5
  37:	c4 62 59 de c5       	vaesdec xmm8,xmm4,xmm5
  3c:	c4 62 59 de cd       	vaesdec xmm9,xmm4,xmm5
  41:	c4 62 59 de d5       	vaesdec xmm10,xmm4,xmm5
  46:	c4 62 59 de dd       	vaesdec xmm11,xmm4,xmm5
  4b:	c4 62 59 de e5       	vaesdec xmm12,xmm4,xmm5
  50:	c4 e2 59 de dd       	vaesdec xmm3,xmm4,xmm5
  55:	c4 e2 59 de f5       	vaesdec xmm6,xmm4,xmm5
  5a:	c4 e2 59 de fd       	vaesdec xmm7,xmm4,xmm5
  5f:	c4 62 59 de c5       	vaesdec xmm8,xmm4,xmm5
  64:	c4 62 59 de cd       	vaesdec xmm9,xmm4,xmm5
  69:	c4 62 59 de d5       	vaesdec xmm10,xmm4,xmm5
  6e:	c4 62 59 de dd       	vaesdec xmm11,xmm4,xmm5
  73:	c4 62 59 de e5       	vaesdec xmm12,xmm4,xmm5
  78:	c4 e2 59 de dd       	vaesdec xmm3,xmm4,xmm5
  7d:	c4 e2 59 de f5       	vaesdec xmm6,xmm4,xmm5
  82:	c4 e2 59 de fd       	vaesdec xmm7,xmm4,xmm5
  87:	c4 62 59 de c5       	vaesdec xmm8,xmm4,xmm5
  8c:	c4 62 59 de cd       	vaesdec xmm9,xmm4,xmm5
  91:	c4 62 59 de d5       	vaesdec xmm10,xmm4,xmm5
  96:	c4 62 59 de dd       	vaesdec xmm11,xmm4,xmm5
  9b:	c4 62 59 de e5       	vaesdec xmm12,xmm4,xmm5
  a0:	62 f2 76 48 52 c2    	vdpbf16ps zmm0,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS XMM4, XMM4, XMM4;
VXORPS XMM5, XMM5, XMM5;
MOV RAX, 0x4000000040000000;
MOV [R14+0], RAX;
MOV [R14+8], RAX;
MOV [R14+16], RAX;
MOV [R14+24], RAX;
MOV [R14+32], RAX;
MOV [R14+40], RAX;
MOV [R14+48], RAX;
MOV [R14+56], RAX;
VMOVUPD ZMM0, [R14];
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 33.0
- Core cycles: 31.08
- Reference cycles: 19.78
- UOPS_PORT_1: 0.9
- UOPS_PORT_0: 31.1

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

Code:

   0:	0f 6f c1             	movq   mm0,mm1
   3:	0f 6f d1             	movq   mm2,mm1
   6:	0f 6f d9             	movq   mm3,mm1
   9:	0f 6f e1             	movq   mm4,mm1
   c:	0f 6f e9             	movq   mm5,mm1
   f:	0f 6f f1             	movq   mm6,mm1
  12:	0f 6f f9             	movq   mm7,mm1
  15:	0f 6f c1             	movq   mm0,mm1
  18:	0f 6f d1             	movq   mm2,mm1
  1b:	0f 6f d9             	movq   mm3,mm1
  1e:	0f 6f e1             	movq   mm4,mm1
  21:	0f 6f e9             	movq   mm5,mm1
  24:	0f 6f f1             	movq   mm6,mm1
  27:	0f 6f f9             	movq   mm7,mm1
  2a:	0f 6f c1             	movq   mm0,mm1
  2d:	0f 6f d1             	movq   mm2,mm1
  30:	0f 6f d9             	movq   mm3,mm1
  33:	0f 6f e1             	movq   mm4,mm1
  36:	0f 6f e9             	movq   mm5,mm1
  39:	0f 6f f1             	movq   mm6,mm1
  3c:	0f 6f f9             	movq   mm7,mm1
  3f:	0f 6f c1             	movq   mm0,mm1
  42:	0f 6f d1             	movq   mm2,mm1
  45:	0f 6f d9             	movq   mm3,mm1
  48:	0f 6f e1             	movq   mm4,mm1
  4b:	0f 6f e9             	movq   mm5,mm1
  4e:	0f 6f f1             	movq   mm6,mm1
  51:	0f 6f f9             	movq   mm7,mm1
  54:	0f 6f c1             	movq   mm0,mm1
  57:	0f 6f d1             	movq   mm2,mm1
  5a:	0f 6f d9             	movq   mm3,mm1
  5d:	0f 6f e1             	movq   mm4,mm1
  60:	62 f2 76 48 52 c2    	vdpbf16ps zmm0,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
PXOR MM1, MM1;
VZEROALL;
MOV RAX, 0x4000000040000000;
MOV [R14+0], RAX;
MOV [R14+8], RAX;
MOV [R14+16], RAX;
MOV [R14+24], RAX;
MOV [R14+32], RAX;
MOV [R14+40], RAX;
MOV [R14+48], RAX;
MOV [R14+56], RAX;
VMOVUPD ZMM0, [R14];
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 33.0
- Core cycles: 18.22
- Reference cycles: 12.47
- UOPS_PORT_0: 17.94
- UOPS_PORT_5B: 18.06
- UOPS_PORT_5B>=2: 0.0

⇨ 2 μops that can only use ports {'0', '5'}