VGATHERDPS (ZMM, K, VSIB

With unroll_count=500 and no inner loop

Code:

   0:	62 92 7d 49 92 04 36 	vgatherdps zmm0{k1},DWORD PTR [r14+zmm14*1]

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 ZMM14, ZMM14, ZMM14

Show nanoBench command
Results:
- MPERF: 14.21
- APERF: 16.49
- UOPS: 81.0
- FpuPipeAssignment.Total0: 6.28
- FpuPipeAssignment.Total1: 5.2
- FpuPipeAssignment.Total2: 5.04
- FpuPipeAssignment.Total3: 5.49
- FpuPipeAssignment.Total4: 9.19
- FpuPipeAssignment.Total5: 8.76
- DIV_CYCLES: 0.0

With loop_count=1000 and unroll_count=10

Code:

   0:	62 92 7d 49 92 04 36 	vgatherdps zmm0{k1},DWORD PTR [r14+zmm14*1]

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 ZMM14, ZMM14, ZMM14

Show nanoBench command
Results:
- MPERF: 14.3
- APERF: 16.59
- UOPS: 81.1
- FpuPipeAssignment.Total0: 6.29
- FpuPipeAssignment.Total1: 5.26
- FpuPipeAssignment.Total2: 5.25
- FpuPipeAssignment.Total3: 5.2
- FpuPipeAssignment.Total4: 9.01
- FpuPipeAssignment.Total5: 9.02
- DIV_CYCLES: 0.0

With loop_count=100 and unroll_count=100

Code:

   0:	62 92 7d 49 92 04 36 	vgatherdps zmm0{k1},DWORD PTR [r14+zmm14*1]

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 ZMM14, ZMM14, ZMM14

Show nanoBench command
Results:
- MPERF: 14.29
- APERF: 16.57
- UOPS: 81.01
- FpuPipeAssignment.Total0: 6.29
- FpuPipeAssignment.Total1: 5.19
- FpuPipeAssignment.Total2: 5.19
- FpuPipeAssignment.Total3: 5.33
- FpuPipeAssignment.Total4: 8.99
- FpuPipeAssignment.Total5: 9.05
- DIV_CYCLES: 0.0

With 4 independent instructions

With unroll_count=200 and no inner loop

Code:

   0:	62 92 7d 49 92 04 36 	vgatherdps zmm0{k1},DWORD PTR [r14+zmm14*1]
   7:	62 92 7d 4a 92 4c 36 	vgatherdps zmm1{k2},DWORD PTR [r14+zmm14*1+0x4]
   e:	01 
   f:	62 92 7d 4b 92 54 36 	vgatherdps zmm2{k3},DWORD PTR [r14+zmm14*1+0x8]
  16:	02 
  17:	62 92 7d 4c 92 5c 36 	vgatherdps zmm3{k4},DWORD PTR [r14+zmm14*1+0xc]
  1e:	03

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 ZMM14, ZMM14, ZMM14;
VMOVUPD ZMM1, [R14];
MOV [R14+4], RAX;
VMOVUPD ZMM2, [R14];
VMOVUPD ZMM3, [R14];
MOV [R14+12], RAX

Show nanoBench command
Results:
- MPERF: 57.42
- APERF: 66.56
- UOPS: 324.0
- FpuPipeAssignment.Total0: 21.07
- FpuPipeAssignment.Total1: 22.15
- FpuPipeAssignment.Total2: 24.14
- FpuPipeAssignment.Total3: 20.63
- FpuPipeAssignment.Total4: 35.52
- FpuPipeAssignment.Total5: 36.49
- DIV_CYCLES: 0.0

With loop_count=1000 and unroll_count=2

Code:

   0:	62 92 7d 49 92 04 36 	vgatherdps zmm0{k1},DWORD PTR [r14+zmm14*1]
   7:	62 92 7d 4a 92 4c 36 	vgatherdps zmm1{k2},DWORD PTR [r14+zmm14*1+0x4]
   e:	01 
   f:	62 92 7d 4b 92 54 36 	vgatherdps zmm2{k3},DWORD PTR [r14+zmm14*1+0x8]
  16:	02 
  17:	62 92 7d 4c 92 5c 36 	vgatherdps zmm3{k4},DWORD PTR [r14+zmm14*1+0xc]
  1e:	03

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 ZMM14, ZMM14, ZMM14;
VMOVUPD ZMM1, [R14];
MOV [R14+4], RAX;
VMOVUPD ZMM2, [R14];
VMOVUPD ZMM3, [R14];
MOV [R14+12], RAX

Show nanoBench command
Results:
- MPERF: 57.74
- APERF: 66.96
- UOPS: 324.5
- FpuPipeAssignment.Total0: 21.09
- FpuPipeAssignment.Total1: 23.79
- FpuPipeAssignment.Total2: 22.89
- FpuPipeAssignment.Total3: 20.24
- FpuPipeAssignment.Total4: 35.99
- FpuPipeAssignment.Total5: 36.02
- DIV_CYCLES: 0.0

With loop_count=100 and unroll_count=20

Code:

   0:	62 92 7d 49 92 04 36 	vgatherdps zmm0{k1},DWORD PTR [r14+zmm14*1]
   7:	62 92 7d 4a 92 4c 36 	vgatherdps zmm1{k2},DWORD PTR [r14+zmm14*1+0x4]
   e:	01 
   f:	62 92 7d 4b 92 54 36 	vgatherdps zmm2{k3},DWORD PTR [r14+zmm14*1+0x8]
  16:	02 
  17:	62 92 7d 4c 92 5c 36 	vgatherdps zmm3{k4},DWORD PTR [r14+zmm14*1+0xc]
  1e:	03

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 ZMM14, ZMM14, ZMM14;
VMOVUPD ZMM1, [R14];
MOV [R14+4], RAX;
VMOVUPD ZMM2, [R14];
VMOVUPD ZMM3, [R14];
MOV [R14+12], RAX

Show nanoBench command
Results:
- MPERF: 57.58
- APERF: 66.77
- UOPS: 324.05
- FpuPipeAssignment.Total0: 21.44
- FpuPipeAssignment.Total1: 22.19
- FpuPipeAssignment.Total2: 24.35
- FpuPipeAssignment.Total3: 20.02
- FpuPipeAssignment.Total4: 35.07
- FpuPipeAssignment.Total5: 36.93
- DIV_CYCLES: 0.0

With 7 independent instructions

With unroll_count=100 and no inner loop

Code:

   0:	62 92 7d 49 92 04 36 	vgatherdps zmm0{k1},DWORD PTR [r14+zmm14*1]
   7:	62 92 7d 4a 92 4c 36 	vgatherdps zmm1{k2},DWORD PTR [r14+zmm14*1+0x4]
   e:	01 
   f:	62 92 7d 4b 92 54 36 	vgatherdps zmm2{k3},DWORD PTR [r14+zmm14*1+0x8]
  16:	02 
  17:	62 92 7d 4c 92 5c 36 	vgatherdps zmm3{k4},DWORD PTR [r14+zmm14*1+0xc]
  1e:	03 
  1f:	62 92 7d 4d 92 64 36 	vgatherdps zmm4{k5},DWORD PTR [r14+zmm14*1+0x10]
  26:	04 
  27:	62 92 7d 4e 92 6c 36 	vgatherdps zmm5{k6},DWORD PTR [r14+zmm14*1+0x14]
  2e:	05 
  2f:	62 92 7d 4f 92 74 36 	vgatherdps zmm6{k7},DWORD PTR [r14+zmm14*1+0x18]
  36:	06

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 ZMM14, ZMM14, ZMM14;
VMOVUPD ZMM1, [R14];
MOV [R14+4], RAX;
VMOVUPD ZMM2, [R14];
VMOVUPD ZMM3, [R14];
MOV [R14+12], RAX;
VMOVUPD ZMM4, [R14];
VMOVUPD ZMM5, [R14];
MOV [R14+20], RAX;
VMOVUPD ZMM6, [R14]

Show nanoBench command
Results:
- MPERF: 100.27
- APERF: 116.2
- UOPS: 567.0
- FpuPipeAssignment.Total0: 36.62
- FpuPipeAssignment.Total1: 38.7
- FpuPipeAssignment.Total2: 42.04
- FpuPipeAssignment.Total3: 36.65
- FpuPipeAssignment.Total4: 62.67
- FpuPipeAssignment.Total5: 63.65
- DIV_CYCLES: 0.0

With loop_count=1000 and unroll_count=1

Code:

   0:	62 92 7d 49 92 04 36 	vgatherdps zmm0{k1},DWORD PTR [r14+zmm14*1]
   7:	62 92 7d 4a 92 4c 36 	vgatherdps zmm1{k2},DWORD PTR [r14+zmm14*1+0x4]
   e:	01 
   f:	62 92 7d 4b 92 54 36 	vgatherdps zmm2{k3},DWORD PTR [r14+zmm14*1+0x8]
  16:	02 
  17:	62 92 7d 4c 92 5c 36 	vgatherdps zmm3{k4},DWORD PTR [r14+zmm14*1+0xc]
  1e:	03 
  1f:	62 92 7d 4d 92 64 36 	vgatherdps zmm4{k5},DWORD PTR [r14+zmm14*1+0x10]
  26:	04 
  27:	62 92 7d 4e 92 6c 36 	vgatherdps zmm5{k6},DWORD PTR [r14+zmm14*1+0x14]
  2e:	05 
  2f:	62 92 7d 4f 92 74 36 	vgatherdps zmm6{k7},DWORD PTR [r14+zmm14*1+0x18]
  36:	06

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 ZMM14, ZMM14, ZMM14;
VMOVUPD ZMM1, [R14];
MOV [R14+4], RAX;
VMOVUPD ZMM2, [R14];
VMOVUPD ZMM3, [R14];
MOV [R14+12], RAX;
VMOVUPD ZMM4, [R14];
VMOVUPD ZMM5, [R14];
MOV [R14+20], RAX;
VMOVUPD ZMM6, [R14]

Show nanoBench command
Results:
- MPERF: 101.35
- APERF: 117.53
- UOPS: 568.0
- FpuPipeAssignment.Total0: 37.02
- FpuPipeAssignment.Total1: 40.59
- FpuPipeAssignment.Total2: 41.05
- FpuPipeAssignment.Total3: 35.35
- FpuPipeAssignment.Total4: 62.48
- FpuPipeAssignment.Total5: 63.53
- DIV_CYCLES: 0.0

With loop_count=100 and unroll_count=10

Code:

   0:	62 92 7d 49 92 04 36 	vgatherdps zmm0{k1},DWORD PTR [r14+zmm14*1]
   7:	62 92 7d 4a 92 4c 36 	vgatherdps zmm1{k2},DWORD PTR [r14+zmm14*1+0x4]
   e:	01 
   f:	62 92 7d 4b 92 54 36 	vgatherdps zmm2{k3},DWORD PTR [r14+zmm14*1+0x8]
  16:	02 
  17:	62 92 7d 4c 92 5c 36 	vgatherdps zmm3{k4},DWORD PTR [r14+zmm14*1+0xc]
  1e:	03 
  1f:	62 92 7d 4d 92 64 36 	vgatherdps zmm4{k5},DWORD PTR [r14+zmm14*1+0x10]
  26:	04 
  27:	62 92 7d 4e 92 6c 36 	vgatherdps zmm5{k6},DWORD PTR [r14+zmm14*1+0x14]
  2e:	05 
  2f:	62 92 7d 4f 92 74 36 	vgatherdps zmm6{k7},DWORD PTR [r14+zmm14*1+0x18]
  36:	06

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 ZMM14, ZMM14, ZMM14;
VMOVUPD ZMM1, [R14];
MOV [R14+4], RAX;
VMOVUPD ZMM2, [R14];
VMOVUPD ZMM3, [R14];
MOV [R14+12], RAX;
VMOVUPD ZMM4, [R14];
VMOVUPD ZMM5, [R14];
MOV [R14+20], RAX;
VMOVUPD ZMM6, [R14]

Show nanoBench command
Results:
- MPERF: 100.63
- APERF: 116.68
- UOPS: 567.1
- FpuPipeAssignment.Total0: 37.51
- FpuPipeAssignment.Total1: 38.05
- FpuPipeAssignment.Total2: 42.95
- FpuPipeAssignment.Total3: 35.5
- FpuPipeAssignment.Total4: 61.27
- FpuPipeAssignment.Total5: 64.73
- DIV_CYCLES: 0.0

VGATHERDPS (ZMM, K, VSIB_ZMM) - Throughput and Uops

With 1 independent instruction

With unroll_count=500 and no inner loop

With loop_count=1000 and unroll_count=10

With loop_count=100 and unroll_count=100

With 4 independent instructions

With unroll_count=200 and no inner loop

With loop_count=1000 and unroll_count=2

With loop_count=100 and unroll_count=20

With 7 independent instructions

With unroll_count=100 and no inner loop

With loop_count=1000 and unroll_count=1

With loop_count=100 and unroll_count=10