VPERMB (ZMM, ZMM, ZMM)

With unroll_count=500 and no inner loop

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 1.0
- Core cycles: 1.0
- Reference cycles: 0.61
- UOPS_EXECUTED.THREAD: 1.0
- RETIRE_SLOTS: 1.0
- UOPS_MITE: 1.0
- UOPS_MS: 0.0
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 1.0
- UOPS_PORT_6: 0.0
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.25

With loop_count=1000 and unroll_count=10

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 1.2
- Core cycles: 1.0
- Reference cycles: 0.61
- UOPS_EXECUTED.THREAD: 1.1
- RETIRE_SLOTS: 1.1
- UOPS_MITE: 0.0
- UOPS_MS: 0.0
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 1.0
- UOPS_PORT_6: 0.1
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.0

With loop_count=100 and unroll_count=100

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 1.02
- Core cycles: 1.0
- Reference cycles: 0.61
- UOPS_EXECUTED.THREAD: 1.01
- RETIRE_SLOTS: 1.01
- UOPS_MITE: 0.01
- UOPS_MS: 0.0
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 1.0
- UOPS_PORT_6: 0.01
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.0

With 4 independent instructions

With unroll_count=200 and no inner loop

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2
   6:	62 f2 75 48 8d da    	vpermb zmm3,zmm1,zmm2
   c:	62 f2 75 48 8d e2    	vpermb zmm4,zmm1,zmm2
  12:	62 f2 75 48 8d ea    	vpermb zmm5,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 4.0
- Core cycles: 4.0
- Reference cycles: 2.31
- UOPS_EXECUTED.THREAD: 4.0
- RETIRE_SLOTS: 4.0
- UOPS_MITE: 4.0
- UOPS_MS: 0.0
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 4.0
- UOPS_PORT_6: 0.0
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 1.0

With loop_count=1000 and unroll_count=2

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2
   6:	62 f2 75 48 8d da    	vpermb zmm3,zmm1,zmm2
   c:	62 f2 75 48 8d e2    	vpermb zmm4,zmm1,zmm2
  12:	62 f2 75 48 8d ea    	vpermb zmm5,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 5.0
- Core cycles: 4.0
- Reference cycles: 2.45
- UOPS_EXECUTED.THREAD: 4.5
- RETIRE_SLOTS: 4.5
- UOPS_MITE: 0.01
- UOPS_MS: 0.01
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 4.0
- UOPS_PORT_6: 0.5
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.0

With loop_count=100 and unroll_count=20

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2
   6:	62 f2 75 48 8d da    	vpermb zmm3,zmm1,zmm2
   c:	62 f2 75 48 8d e2    	vpermb zmm4,zmm1,zmm2
  12:	62 f2 75 48 8d ea    	vpermb zmm5,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 4.1
- Core cycles: 4.0
- Reference cycles: 2.45
- UOPS_EXECUTED.THREAD: 4.05
- RETIRE_SLOTS: 4.05
- UOPS_MITE: 0.04
- UOPS_MS: 0.01
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 4.0
- UOPS_PORT_6: 0.05
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.01

With 8 independent instructions

With unroll_count=100 and no inner loop

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2
   6:	62 f2 75 48 8d da    	vpermb zmm3,zmm1,zmm2
   c:	62 f2 75 48 8d e2    	vpermb zmm4,zmm1,zmm2
  12:	62 f2 75 48 8d ea    	vpermb zmm5,zmm1,zmm2
  18:	62 f2 75 48 8d f2    	vpermb zmm6,zmm1,zmm2
  1e:	62 f2 75 48 8d fa    	vpermb zmm7,zmm1,zmm2
  24:	62 72 75 48 8d c2    	vpermb zmm8,zmm1,zmm2
  2a:	62 72 75 48 8d ca    	vpermb zmm9,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 8.0
- Core cycles: 8.0
- Reference cycles: 4.87
- UOPS_EXECUTED.THREAD: 8.0
- RETIRE_SLOTS: 8.0
- UOPS_MITE: 8.0
- UOPS_MS: 0.0
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 8.0
- UOPS_PORT_6: 0.0
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 2.0

With loop_count=1000 and unroll_count=1

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2
   6:	62 f2 75 48 8d da    	vpermb zmm3,zmm1,zmm2
   c:	62 f2 75 48 8d e2    	vpermb zmm4,zmm1,zmm2
  12:	62 f2 75 48 8d ea    	vpermb zmm5,zmm1,zmm2
  18:	62 f2 75 48 8d f2    	vpermb zmm6,zmm1,zmm2
  1e:	62 f2 75 48 8d fa    	vpermb zmm7,zmm1,zmm2
  24:	62 72 75 48 8d c2    	vpermb zmm8,zmm1,zmm2
  2a:	62 72 75 48 8d ca    	vpermb zmm9,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 10.0
- Core cycles: 8.01
- Reference cycles: 4.89
- UOPS_EXECUTED.THREAD: 9.0
- RETIRE_SLOTS: 9.0
- UOPS_MITE: 0.01
- UOPS_MS: 0.02
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 8.0
- UOPS_PORT_6: 1.0
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.01

With loop_count=100 and unroll_count=10

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2
   6:	62 f2 75 48 8d da    	vpermb zmm3,zmm1,zmm2
   c:	62 f2 75 48 8d e2    	vpermb zmm4,zmm1,zmm2
  12:	62 f2 75 48 8d ea    	vpermb zmm5,zmm1,zmm2
  18:	62 f2 75 48 8d f2    	vpermb zmm6,zmm1,zmm2
  1e:	62 f2 75 48 8d fa    	vpermb zmm7,zmm1,zmm2
  24:	62 72 75 48 8d c2    	vpermb zmm8,zmm1,zmm2
  2a:	62 72 75 48 8d ca    	vpermb zmm9,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 8.2
- Core cycles: 8.01
- Reference cycles: 4.89
- UOPS_EXECUTED.THREAD: 8.1
- RETIRE_SLOTS: 8.11
- UOPS_MITE: 0.09
- UOPS_MS: 0.02
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 8.0
- UOPS_PORT_6: 0.1
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.02

With 16 independent instructions

With unroll_count=100 and no inner loop

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2
   6:	62 f2 75 48 8d da    	vpermb zmm3,zmm1,zmm2
   c:	62 f2 75 48 8d e2    	vpermb zmm4,zmm1,zmm2
  12:	62 f2 75 48 8d ea    	vpermb zmm5,zmm1,zmm2
  18:	62 f2 75 48 8d f2    	vpermb zmm6,zmm1,zmm2
  1e:	62 f2 75 48 8d fa    	vpermb zmm7,zmm1,zmm2
  24:	62 72 75 48 8d c2    	vpermb zmm8,zmm1,zmm2
  2a:	62 72 75 48 8d ca    	vpermb zmm9,zmm1,zmm2
  30:	62 72 75 48 8d d2    	vpermb zmm10,zmm1,zmm2
  36:	62 72 75 48 8d da    	vpermb zmm11,zmm1,zmm2
  3c:	62 72 75 48 8d e2    	vpermb zmm12,zmm1,zmm2
  42:	62 e2 75 48 8d c2    	vpermb zmm16,zmm1,zmm2
  48:	62 e2 75 48 8d ca    	vpermb zmm17,zmm1,zmm2
  4e:	62 e2 75 48 8d d2    	vpermb zmm18,zmm1,zmm2
  54:	62 e2 75 48 8d da    	vpermb zmm19,zmm1,zmm2
  5a:	62 e2 75 48 8d e2    	vpermb zmm20,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 16.0
- Core cycles: 16.0
- Reference cycles: 10.1
- UOPS_EXECUTED.THREAD: 16.0
- RETIRE_SLOTS: 16.0
- UOPS_MITE: 16.0
- UOPS_MS: 0.0
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 16.0
- UOPS_PORT_6: 0.0
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 4.0

With loop_count=1000 and unroll_count=1

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2
   6:	62 f2 75 48 8d da    	vpermb zmm3,zmm1,zmm2
   c:	62 f2 75 48 8d e2    	vpermb zmm4,zmm1,zmm2
  12:	62 f2 75 48 8d ea    	vpermb zmm5,zmm1,zmm2
  18:	62 f2 75 48 8d f2    	vpermb zmm6,zmm1,zmm2
  1e:	62 f2 75 48 8d fa    	vpermb zmm7,zmm1,zmm2
  24:	62 72 75 48 8d c2    	vpermb zmm8,zmm1,zmm2
  2a:	62 72 75 48 8d ca    	vpermb zmm9,zmm1,zmm2
  30:	62 72 75 48 8d d2    	vpermb zmm10,zmm1,zmm2
  36:	62 72 75 48 8d da    	vpermb zmm11,zmm1,zmm2
  3c:	62 72 75 48 8d e2    	vpermb zmm12,zmm1,zmm2
  42:	62 e2 75 48 8d c2    	vpermb zmm16,zmm1,zmm2
  48:	62 e2 75 48 8d ca    	vpermb zmm17,zmm1,zmm2
  4e:	62 e2 75 48 8d d2    	vpermb zmm18,zmm1,zmm2
  54:	62 e2 75 48 8d da    	vpermb zmm19,zmm1,zmm2
  5a:	62 e2 75 48 8d e2    	vpermb zmm20,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 18.0
- Core cycles: 16.01
- Reference cycles: 9.78
- UOPS_EXECUTED.THREAD: 17.0
- RETIRE_SLOTS: 17.01
- UOPS_MITE: 0.02
- UOPS_MS: 0.02
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 16.0
- UOPS_PORT_6: 1.0
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.0

With loop_count=100 and unroll_count=10

Code:

   0:	62 f2 75 48 8d c2    	vpermb zmm0,zmm1,zmm2
   6:	62 f2 75 48 8d da    	vpermb zmm3,zmm1,zmm2
   c:	62 f2 75 48 8d e2    	vpermb zmm4,zmm1,zmm2
  12:	62 f2 75 48 8d ea    	vpermb zmm5,zmm1,zmm2
  18:	62 f2 75 48 8d f2    	vpermb zmm6,zmm1,zmm2
  1e:	62 f2 75 48 8d fa    	vpermb zmm7,zmm1,zmm2
  24:	62 72 75 48 8d c2    	vpermb zmm8,zmm1,zmm2
  2a:	62 72 75 48 8d ca    	vpermb zmm9,zmm1,zmm2
  30:	62 72 75 48 8d d2    	vpermb zmm10,zmm1,zmm2
  36:	62 72 75 48 8d da    	vpermb zmm11,zmm1,zmm2
  3c:	62 72 75 48 8d e2    	vpermb zmm12,zmm1,zmm2
  42:	62 e2 75 48 8d c2    	vpermb zmm16,zmm1,zmm2
  48:	62 e2 75 48 8d ca    	vpermb zmm17,zmm1,zmm2
  4e:	62 e2 75 48 8d d2    	vpermb zmm18,zmm1,zmm2
  54:	62 e2 75 48 8d da    	vpermb zmm19,zmm1,zmm2
  5a:	62 e2 75 48 8d e2    	vpermb zmm20,zmm1,zmm2

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM1, ZMM1, ZMM1;
VXORPS ZMM2, ZMM2, ZMM2

Show nanoBench command
Results:
- Instructions retired: 16.2
- Core cycles: 16.01
- Reference cycles: 9.78
- UOPS_EXECUTED.THREAD: 16.1
- RETIRE_SLOTS: 16.1
- UOPS_MITE: 0.18
- UOPS_MS: 0.02
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 16.0
- UOPS_PORT_6: 0.1
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.05

With the same register for for different operands

With unroll_count=500 and no inner loop

Code:

   0:	62 f2 7d 48 8d c0    	vpermb zmm0,zmm0,zmm0

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM0, ZMM0, ZMM0

Show nanoBench command
Results:
- Instructions retired: 1.0
- Core cycles: 3.0
- Reference cycles: 1.8
- UOPS_EXECUTED.THREAD: 1.0
- RETIRE_SLOTS: 1.0
- UOPS_MITE: 0.95
- UOPS_MS: 0.0
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 1.0
- UOPS_PORT_6: 0.0
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.25

With loop_count=1000 and unroll_count=10

Code:

   0:	62 f2 7d 48 8d c0    	vpermb zmm0,zmm0,zmm0

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM0, ZMM0, ZMM0

Show nanoBench command
Results:
- Instructions retired: 1.2
- Core cycles: 3.0
- Reference cycles: 1.83
- UOPS_EXECUTED.THREAD: 1.1
- RETIRE_SLOTS: 1.1
- UOPS_MITE: 0.0
- UOPS_MS: 0.0
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 1.0
- UOPS_PORT_6: 0.1
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.0

With loop_count=100 and unroll_count=100

Code:

   0:	62 f2 7d 48 8d c0    	vpermb zmm0,zmm0,zmm0

Init:

MOV R15, 10000;
L: VADDPS ZMM0, ZMM1, ZMM1;
VADDPS ZMM0, ZMM1, ZMM1;
DEC R15;
JNZ L;
VZEROALL;
VXORPS ZMM0, ZMM0, ZMM0

Show nanoBench command
Results:
- Instructions retired: 1.02
- Core cycles: 3.0
- Reference cycles: 1.83
- UOPS_EXECUTED.THREAD: 1.01
- RETIRE_SLOTS: 1.01
- UOPS_MITE: 0.01
- UOPS_MS: 0.0
- UOPS_PORT_0: 0.0
- UOPS_PORT_1: 0.0
- UOPS_PORT_5: 1.0
- UOPS_PORT_6: 0.01
- UOPS_PORT_23: 0.0
- UOPS_PORT_49: 0.0
- UOPS_PORT_78: 0.0
- DIV_CYCLES: 0.0
- ILD_STALL.LCP: 0.0
- UOPS_MITE>=1: 0.0

VPERMB (ZMM, ZMM, ZMM) - Throughput and Uops

With different registers for different operands

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 8 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

With 16 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

With the same register for for different operands

With unroll_count=500 and no inner loop

With loop_count=1000 and unroll_count=10

With loop_count=100 and unroll_count=100