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

SUQADD (vector, 8H)

Test 1: uops

Code:

  suqadd v0.8h, v1.8h
  movi v0.16b, 1
  movi v1.16b, 2

(no loop instructions)

1000 unrolls and 1 iteration

Retires: 1.000

Issues: 1.000

Integer unit issues: 0.001

Load/store unit issues: 0.000

SIMD/FP unit issues: 1.000

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)dispatch simd uop (57)ldst uops in schedulers (5b)dispatch uop (78)map simd uop (7e)map simd uop inputs (81)? int output thing (e9)? simd retires (ee)
1004303310011100010007590510001000200011000
1004303310011100010007590510001000200011000
1004303310011100010007590510001000200011000
1004303310011100010007590510001000200011000
1004303310011100010007590510001000200011000
1004303310011100010007590510001000200011000
1004303310011100010007590510001000200011000
1004303310011100010007590510001000200011000
1004303310011100010007590510001000200011000
1004303310011100010007590510001000200011000

Test 2: Latency 1->1

Code:

  suqadd v0.8h, v1.8h
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0033

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)dispatch int uop (56)dispatch simd uop (57)int uops in schedulers (59)ldst uops in schedulers (5b)dispatch uop (78)map int uop (7c)map simd uop (7e)map int uop inputs (7f)map simd uop inputs (81)? int output thing (e9)? simd retires (ee)? int retires (ef)
1020430033101011011000010010000300768905101002001000620020012110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420220092110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300771089102322001016020020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0033

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)dispatch int uop (56)dispatch simd uop (57)int uops in schedulers (59)ldst uops in schedulers (5b)dispatch uop (78)map int uop (7c)map simd uop (7e)map int uop inputs (7f)map simd uop inputs (81)? int output thing (e9)? simd retires (ee)? int retires (ef)
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010

Test 3: Latency 1->2

Code:

  suqadd v0.8h, v0.8h
  movi v0.16b, 1

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0033

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)dispatch int uop (56)dispatch simd uop (57)int uops in schedulers (59)ldst uops in schedulers (5b)dispatch uop (78)map int uop (7c)map simd uop (7e)map int uop inputs (7f)map simd uop inputs (81)? int output thing (e9)? simd retires (ee)? int retires (ef)
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420220092210000100
1020430206101171011001610010066307769366101352021004220020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000314769997101702041008420020008110000100

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0033

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)dispatch int uop (56)dispatch simd uop (57)int uops in schedulers (59)ldst uops in schedulers (5b)dispatch uop (78)map int uop (7c)map simd uop (7e)map int uop inputs (7f)map simd uop inputs (81)? int output thing (e9)? simd retires (ee)? int retires (ef)
10024300331002121100002010000707689051002020100042020008111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100062020000111000010
10024300331002121100002010000707689051002020100002020000111000010

Test 4: throughput

Count: 8

Code:

  movi v0.16b, 0
  suqadd v0.8h, v8.8h
  movi v1.16b, 0
  suqadd v1.8h, v8.8h
  movi v2.16b, 0
  suqadd v2.8h, v8.8h
  movi v3.16b, 0
  suqadd v3.8h, v8.8h
  movi v4.16b, 0
  suqadd v4.8h, v8.8h
  movi v5.16b, 0
  suqadd v5.8h, v8.8h
  movi v6.16b, 0
  suqadd v6.8h, v8.8h
  movi v7.16b, 0
  suqadd v7.8h, v8.8h
  movi v8.16b, 9

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.5011

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)dispatch int uop (56)dispatch simd uop (57)int uops in schedulers (59)ldst uops in schedulers (5b)dispatch uop (78)map int uop (7c)map simd uop (7e)map int uop inputs (7f)map simd uop inputs (81)? int output thing (e9)? simd retires (ee)? int retires (ef)
1602044052180109101800081008001230032005280112200800122001600261160000100
1602044010880110101800091008001330032005280112200800122001600241160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044009680110101800091008001330032005280112200800122001600241160000100
1602044011180110101800091008001330032005280112200800122001600241160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044010880109101800081008001230032005280112200800122001600241160000100

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.5053

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)dispatch int uop (56)dispatch simd uop (57)int uops in schedulers (59)ldst uops in schedulers (5b)dispatch uop (78)map int uop (7c)map simd uop (7e)map int uop inputs (7f)map simd uop inputs (81)? int output thing (e9)? simd retires (ee)? int retires (ef)
1600244382080019118000810800123032005280022208001220160026116000010
1600244122980011118000010800003032000080010208000020160000116000010
1600244041380011118000010800003032000080010208000020160000116000010
1600244041480011118000010800003032000080010208000020160000116000010
1600244041880011118000010800003032000080010208000020160000116000010
1600244042380011118000010800003032000080010208000020160000116000010
1600244038480011118000010800003032000080010208000020160000116000010
1600244042380011118000010800003032000080010208000020160000116000010
1600244042080011118000010800003032000080010208000020160000116000010
1600244042080011118000010800003032000080010208000020160000116000010

Test 5: throughput

Count: 16

Code:

  suqadd v0.8h, v16.8h
  suqadd v1.8h, v16.8h
  suqadd v2.8h, v16.8h
  suqadd v3.8h, v16.8h
  suqadd v4.8h, v16.8h
  suqadd v5.8h, v16.8h
  suqadd v6.8h, v16.8h
  suqadd v7.8h, v16.8h
  suqadd v8.8h, v16.8h
  suqadd v9.8h, v16.8h
  suqadd v10.8h, v16.8h
  suqadd v11.8h, v16.8h
  suqadd v12.8h, v16.8h
  suqadd v13.8h, v16.8h
  suqadd v14.8h, v16.8h
  suqadd v15.8h, v16.8h
  movi v16.16b, 17

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.5002

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)dispatch int uop (56)dispatch simd uop (57)dispatch ldst uop (58)int uops in schedulers (59)simd uops in schedulers (5a)ldst uops in schedulers (5b)dispatch uop (78)map int uop (7c)map ldst uop (7d)map simd uop (7e)map int uop inputs (7f)map simd uop inputs (81)? int output thing (e9)? simd retires (ee)? int retires (ef)
160204800651601051011600041001600080300064003616010820001600122003200241160000100
160204800351601071011600061001600100300064003616010820001600122003200241160000100
160204800351601051011600041001600080300064003616010820001600122003200241160000100
160204800351601051011600041001600080300064003616010820001600122003200241160000100
160205800711601451011600441001600560300064003616010820001600122003200241160000100
160204800351601051011600041001600080300064003616010820001600122003201201160000100
160204800351601051011600041001600080300064003616010820001600122003200241160000100
160204800351601051011600041001600080300064003616010820001600122003200241160000100
160204800351601051011600041001600080300064003616010820001600122003200241160000100
1602048003516010710116000610016001021304069624499640451164732370525431600852003200241160000100

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.5002

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule simd uop (54)dispatch int uop (56)dispatch simd uop (57)int uops in schedulers (59)ldst uops in schedulers (5b)dispatch uop (78)map int uop (7c)map simd uop (7e)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)
16002480341160015111600041016000830640036160018201600122003200001016000010
16002480035160011111600001016000030640000160010201600002003200001016000010
16002480035160011111600001016000030640200160062201600622003200001016000010
16002480035160011111600001016000030640000160010201600002003200001016000010
16002480035160011111600001016000030640000160010201600002003200001016000010
16002480035160011111600001016000030640000160010201600002003200001016000010
16002480035160011111600001016000030640000160010201600002003200001016000010
16002480035160011111600001016000030640208160064201600632003200001016000010
16002480035160011111600001016000030640000160010201600002003200001016000010
16002480038160011111600001016000030640000160010201600002003200001016000010