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

MLS (by element, 4H)

Test 1: uops

Code:

  mls v0.4h, v1.4h, v2.h[1]
  movi v0.16b, 1
  movi v1.16b, 2
  movi v2.16b, 3

(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)
1004303310011100010007590510001000300011000
1004303310011100010007590510001000300011000
1004303310011100010007590510001000300011000
1004303310011100010007590510001000300011000
1004303310011100010007590510001000300011000
1004303310011100010007590510001000300011000
1004303310011100010007590510001000300011000
1004303310011100010007590510001000300011000
1004303310011100010007590510001000300011000
1004303310011100010007590510001000300011000

Test 2: Latency 1->1

Code:

  mls v0.4h, v1.4h, v2.h[1]
  movi v0.16b, 1
  movi v1.16b, 2
  movi v2.16b, 3

(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)
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430186101271031002410210099300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100

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)
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010

Test 3: Latency 1->2

Code:

  mls v0.4h, v0.4h, v1.h[1]
  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)
1020430033101011011000010010000300768905101002001000620030018110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100

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)
10025300661002721100062010031707689051002020100042030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707689051002020100002030000111000010
10024300331002121100002010000707692471005120100442030000111000010

Test 4: Latency 1->3

Code:

  mls v0.4h, v1.4h, v0.h[1]
  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)
1020430033101011011000010010000300768905101002001000620030018110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100
1020430033101011011000010010000300768905101002001000420030012110000100

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)schedule ldst uop (55)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)
100243003310021211000002010000707689051002020100062030000111000010
100243003310021211000002010000707689051002020100002030000111000010
100243003310021211000002010000707689051002020100002030000111000010
100243003310021211000002010000707689051002020100002030000111000010
100243003310021211000002010000707689051002020100002030000111000010
100243003310021211000002010000707689051002020100002030000111000010
100243003310021211000002010000707689051002020100002030000111000010
100243003310021211000002010000707689051002020100002030000111000010
100243003310021211000002010000707689051002020100002030000111000010
100243003310021211000002010000707689051002020100002030000111000010

Test 5: throughput

Count: 8

Code:

  movi v0.16b, 0
  mls v0.4h, v8.4h, v9.h[1]
  movi v1.16b, 0
  mls v1.4h, v8.4h, v9.h[1]
  movi v2.16b, 0
  mls v2.4h, v8.4h, v9.h[1]
  movi v3.16b, 0
  mls v3.4h, v8.4h, v9.h[1]
  movi v4.16b, 0
  mls v4.4h, v8.4h, v9.h[1]
  movi v5.16b, 0
  mls v5.4h, v8.4h, v9.h[1]
  movi v6.16b, 0
  mls v6.4h, v8.4h, v9.h[1]
  movi v7.16b, 0
  mls v7.4h, v8.4h, v9.h[1]
  movi v8.16b, 9
  movi v9.16b, 10

(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 ldst uop inputs (80)map simd uop inputs (81)? int output thing (e9)? ldst retires (ed)? simd retires (ee)? int retires (ef)
160204404618010910180008100800123003200568011320080013200024003910160000100
160204402068014510180044100800483003200528011220080012200024003610160000100
160204400868010910180008100800123003200528011220080012200024003610160000100
160204400868010910180008100800123003200568011320080013200024003610160000100
160204400868010910180008100800123003200568011320080013200024003610160000100
160204400868010910180008100800123003200528011220080012200024003610160000100
160204400868010910180008100800123003200528011220080012200024003610160000100
160204400868010910180008100800123003200528011220080012200024003610160000100
160204401098011010180009100800133003200568011320080013200024003610160000100
160204400868010910180008100800123003200528011220080012200024003610160000100

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)schedule ldst uop (55)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)
16002443796800191180008010800123032005680023208001320240000116000010
16002441154800111180000010800003032000080010208000020240000116000010
16002440403800111180000010800003032000080010208000020240000116000010
16002440418800111180000010800003032000080010208000020240000116000010
16002440388800111180000010800003032018880056208004620240000116000010
16002440403800111180000010800003032000080010208000020240000116000010
16002440418800111180000010800003032000080010208000020240000116000010
16002440409800111180000010800003032000080010208000020240000116000010
16002440402800111180000010800003032000080010208000020240000116000010
16002440418800111180000010800003032000080010208000020240000116000010

Test 6: throughput

Count: 12

Code:

  mls v0.4h, v12.4h, v13.h[1]
  mls v1.4h, v12.4h, v13.h[1]
  mls v2.4h, v12.4h, v13.h[1]
  mls v3.4h, v12.4h, v13.h[1]
  mls v4.4h, v12.4h, v13.h[1]
  mls v5.4h, v12.4h, v13.h[1]
  mls v6.4h, v12.4h, v13.h[1]
  mls v7.4h, v12.4h, v13.h[1]
  mls v8.4h, v12.4h, v13.h[1]
  mls v9.4h, v12.4h, v13.h[1]
  mls v10.4h, v12.4h, v13.h[1]
  mls v11.4h, v12.4h, v13.h[1]
  movi v12.16b, 13
  movi v13.16b, 14

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.5003

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)
120204600551201061011200051001200090300048004412011020001200142003600421120000100
120204600351201071011200061001200100300048003612010820001200122003600361120000100
120204600351201051011200041001200080300048003612010820001200122003600361120000100
120204600351201051011200041001200080300048003612010820001200122003603481120000100
120204600351201071011200061001200100300048003612010820001200122003600361120000100
120205600701201411011200401001200520300048003612010820001200122003600391120000100
120204600541201071011200061001200100300048003612010820001200122003600361120000100
120204600351201051011200041001200080300048003612010820001200122003600361120000100
120204600351201051011200041001200080300048003612010820001200122003600361120000100
120204600351201051011200041001200080300048003612010820001200122003600361120000100

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.5003

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)
120024602111200151112000410120008030048004012001920012001320360000112000010
120024600421200111112000010120000030048000012001020012000020360000112000010
120024600351200111112000010120000030048000012001020012000020360000112000010
120024600351200111112000010120000030048021612006620012006320360000112000010
120024600351200111112000010120000030048000012001020012000020360000112000010
120024600351200111112000010120000030048000012001020012000020360000112000010
120024600351200111112000010120000030048000012001020012000020360000112000010
120024600351200111112000010120000030048000012001020012000020360000112000010
120024600351200111112000010120000030048000012001020012000020360000112000010
120024600351200111112000010120000030048000012001020012000020360000112000010