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

SUBS (immediate, 64-bit)

Test 1: uops

Code:

  subs x0, x0, #3
  mov x0, 1
  mov x1, 2

(no loop instructions)

1000 unrolls and 1 iteration

Retires: 1.000

Issues: 1.000

Integer unit issues: 1.001

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)? int retires (ef)
100410301001100110002504310001000100010011000
100410301001100110002504310001000100010011000
100410301001100110002504310001000100010011000
100410301001100110002504310001000100010011000
100410301001100110002504310001000100010011000
100410301001100110002504310001000100010011000
100410301001100110002504310001000100010011000
100410301001100110002504310001000100010011000
100410301001100110002504310001000100010011000
100410301001100110002504310001000100010011000

Test 2: Latency 1->2

Code:

  subs x0, x0, #3
  mov x0, 1
  mov x1, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 1.0030

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)
10204100301010110101101082514691010810210102101000110100
10204100301010110101101082516761010910210102081000110100
10204100301010110101101082517741010810208102081000110100
10204100301010110101101082517741010810208102081000110100
10204100301010110101101082517741010810208102081000110100
10204100301010110101101082517741010810208102081000110100
10204100301010110101101082517741010810208102081000110100
10204100301010110101101082517741010810208102081000110100
10204100301010110101101082517741010810208102081000110100
10204100301010110101101082517741010810208102081000110100

1000 unrolls and 10 iterations

Result (median cycles for code): 1.0030

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)
10024100301002110021100282532561002010020100201001110010
10024100301002110021100202531611002010020100201001110010
10024100301002110021100202531611002010020100201001110010
10024100301002110021100202531611002010020100201001110010
10024100301002110021100202531611002010020100201001110010
10024100301002110021100202531611002010020100201001110010
10024100301002110021100202531611002010020100201001110010
10024100301002110021100202531611002010020100201001110010
10024100301002110021100202531611002010020100201001110010
10024100301002110021100202531611002010020100201001110010

Test 3: Latency 3->2

Chain cycles: 1

Code:

  subs x0, x1, #3
  cset x1, cc
  mov x0, 1
  mov x1, 2
  mov x2, 3
  mov x3, 4
  mov x4, 5

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code, minus 1 chain cycle): 1.0030

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)
20204200302010120101201075195482010820216202162000120100
20204200302010120101201085195482010820216202162000120100
20204200302010120101201085195482010820216202162000120100
20204200302010120101201085195482010820216202162000120100
20204200302010120101201085195482010820216202162000120100
20204200302010120101201085195482010820216202162000120100
20204200302010120101201085195482010820216202162000120100
20204200302010120101201085195482010820216202162000120100
20204200302010120101201085195482010820216202162000120100
20204200302010120101201085195482010820216202162000120100

1000 unrolls and 10 iterations

Result (median cycles for code, minus 1 chain cycle): 1.0030

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)map simd uop inputs (81)? int output thing (e9)? ldst retires (ed)? simd retires (ee)? int retires (ef)
200242003020011200112001851945420010200202002000200010020010
200242003020011200112001051959820010200202002000200010020010
200242003020011200112001051959820010200202002000200010020010
200242003020011200112001051967220054200662029900201270020010
200242003020011200112001051959820010200202002000200010020010
200242003020011200112001051959820010200202002000200010020010
200242003020011200112001051959820010200202002000200010020010
200242003020011200112001051959820010200202002000200010020010
200242003020011200112001051959820010200202002000200010020010
200242003020011200112001051959820010200202002000200010020010

Test 4: throughput

Count: 8

Code:

  subs x0, x8, #3
  subs x1, x8, #3
  subs x2, x8, #3
  subs x3, x8, #3
  subs x4, x8, #3
  subs x5, x8, #3
  subs x6, x8, #3
  subs x7, x8, #3
  mov x8, 9
  mov x9, 10
  mov x10, 11

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.5010

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)schedule ldst uop (55)dispatch int uop (56)dispatch simd uop (57)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)map ldst uop inputs (80)map simd uop inputs (81)? int output thing (e9)? ldst retires (ed)? simd retires (ee)? int retires (ef)
802044011180109801090080112024034580115802168021600800120080100
802044008180112801120080115024034580115802168021600800120080100
802044008180112801120080115024034580115802168021600800120080100
8020440081801128011200801150240345801158021684576255415820171077582236
802044008180112801120080115024043580145802458024800800400080100
94199650179449389713454735896173624034580115802168021600800120080100
802044008180112801120080115024034580115802168021600800120080100
802044008180112801120080115024034580115802168021600800120080100
802054010680140801400080147024034580115802168021600800120080100
802044008180112801120080115024034580115802168021600800120080100

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.5004

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)map simd uop inputs (81)? int output thing (e9)? ldst retires (ed)? simd retires (ee)? int retires (ef)
800244016180031800318003424011080035800368002000800110080010
800244003080021800218002024006580020800208002000800110080010
800244003080021800218002024006580020800208002000800110080010
800254006280061800618006824006580020800208002000800110080010
800244003080021800218002024006580020800208002000800110080010
800244003080021800218002024006580020800208002000800110080010
800244003080021800218002024006580020800208002000800110080010
800244003080021800218002024006580020800208002000800110080010
800244003080021800218002024006580020800208002000800110080010
800244003080021800218002024006580020800208002000800110080010