Apple Microarchitecture Research by Dougall Johnson

M1/A14 P-core (Firestorm): Overview | Base Instructions | SIMD and FP Instructions
M1/A14 E-core (Icestorm):  Overview | Base Instructions | SIMD and FP Instructions

CMP (register, lsl, 32-bit)

Test 1: uops

Code:

  cmp w0, w1, lsl #17
  mov x0, 1
  mov x1, 2

(no loop instructions)

1000 unrolls and 1 iteration

Retires: 1.000

Issues: 2.000

Integer unit issues: 2.001

Load/store unit issues: 0.000

SIMD/FP unit issues: 0.000

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)
1004698200120011000146781000100020002001
1004698200120011000146781000100020002001
1004698200120011000146781000100020002001
1004698200120011000146781000100020002001
1004698200120011000146781000100020002001
1004698200120011000146781000100020002001
1004698200120011000146781000100020002001
1004698200120011000146781000100020002001
1004698200120011000146781000100020002001
1004698200120011000146781000100020002001

Test 2: Latency 3->1

Chain cycles: 1

Code:

  cmp w0, w1, lsl #17
  cset x0, cc
  mov x0, 1
  mov x1, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code, minus 1 chain cycle): 2.0030

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)map ldst uop inputs (80)? int output thing (e9)? ldst retires (ed)? simd retires (ee)? int retires (ef)
20204300303010130101201057892472010820214302180300010010100
20204300303010130101201057893692010520212302180300010010100
20204300303010130101201057893692010520212302180300010010100
20204300303010130101201057893692010520212302180300010010100
20204300303010130101201057893692010520212302180300010010100
20204300303010130101201057896992014220256302180300010010100
20204300303010130101201057893692010520212302180300010010100
20204300303010130101201057893692010520212302180300010010100
20204300303010130101201057893692010520212302180300010010100
20204300303010130101201057893692010520212302180300010010100

1000 unrolls and 10 iterations

Result (median cycles for code, minus 1 chain cycle): 2.0030

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
20024300303001130011200157894012001520030300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020302703008510010

Test 3: Latency 3->2

Chain cycles: 1

Code:

  cmp w0, w1, lsl #17
  cset x1, cc
  mov x0, 1
  mov x1, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code, minus 1 chain cycle): 2.0030

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
20204300303010130101201067891722010520210302183000110100
20204300303010130101201057893692010520212302183000110100
20204300303010130101201057892872010520210302183000110100
20204300303010130101201057893692010520212302183000110100
20204300303010130101201057893692010520212302183000110100
20204300303010130101201057893692010520212302183000110100
20204300303010130101201057893692010520212302183000110100
20204300303010130101201057893692010520212302183000110100
20204300303010130101201057893692010520212302183000110100
20204300303010130101201057893692010520212302183000110100

1000 unrolls and 10 iterations

Result (median cycles for code, minus 1 chain cycle): 2.0030

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
20024300303001130011200157893802001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107898042005120076300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010
20024300303001130011200107894382001020020300203000110010

Test 4: throughput

Count: 8

Code:

  cmp w0, w1, lsl #17
  cmp w0, w1, lsl #17
  cmp w0, w1, lsl #17
  cmp w0, w1, lsl #17
  cmp w0, w1, lsl #17
  cmp w0, w1, lsl #17
  cmp w0, w1, lsl #17
  cmp w0, w1, lsl #17
  mov x0, 1
  mov x1, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.6675

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
80204534041601171601178012711776018013080230160252160016100
80204534141601161601168012511776358012380224160248160014100
80204534021601141601148012311776198012380224160248160014100
80204534021601141601148012311776558012380224160248160014100
80204534021601141601148012311776238012380224160248160014100
80204534021601141601148012311776078012380224160248160014100
80204534021601141601148012311776318012380224160248160014100
80204534021601141601148012311780468016380264160248160014100
80204534021601141601148012311776198012380224160336160065100
80204534021601141601148012311776238012380224160248160014100

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.6671

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
8002453401160039160039800471170032800208002016002016001110
8002453371160021160021800201170032800208002016002016001110
8002453371160021160021800201170032800208002016002016001110
8002453371160021160021800201175380800928009216002016001110
8002453371160021160021800201170032800208002016002016001110
8002453371160021160021800201170032800208002016002016001110
8002453371160021160021800201170032800208002016002016001110
8002453371160021160021800201170032800208002016002016001110
8002453371160021160021800201170032800208002016002016001110
8002453371160021160021800201170032800208002016002016001110