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

USRA (vector, 4H)

Test 1: uops

Code:

  usra v0.4h, v1.4h, #3
  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:

  usra v0.4h, v1.4h, #3
  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)
1020430033101011011000010010000300768905101002001000620020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
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)
10024300331002121100002010000707689051002020100062020000111000010
10024300331002121100002010000677692471005120100472020088111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010
10024300331002121100002010000707689051002020100002020000111000010

Test 3: Latency 1->2

Code:

  usra v0.4h, v0.4h, #3
  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
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
1020430033101011011000010010000300768905101002001000420020008110000100
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 4: throughput

Count: 8

Code:

  movi v0.16b, 0
  usra v0.4h, v8.4h, #3
  movi v1.16b, 0
  usra v1.4h, v8.4h, #3
  movi v2.16b, 0
  usra v2.4h, v8.4h, #3
  movi v3.16b, 0
  usra v3.4h, v8.4h, #3
  movi v4.16b, 0
  usra v4.4h, v8.4h, #3
  movi v5.16b, 0
  usra v5.4h, v8.4h, #3
  movi v6.16b, 0
  usra v6.4h, v8.4h, #3
  movi v7.16b, 0
  usra v7.4h, v8.4h, #3
  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)
1602044046880109101800081008001230032005280112200800122001600241160000100
1602044009980110101800091008001330032005680113200800132001600241160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044008680109101800081008001230032020480150200800502001600261160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044008680109101800081008001230032005680113200800132001600241160000100
1602044008680109101800081008001230032005280112200800122001600241160000100
1602044009680110101800091008001330032005680113200800132001600241160000100

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)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 ldst uop inputs (80)map simd uop inputs (81)? int output thing (e9)? ldst retires (ed)? simd retires (ee)? int retires (ef)
160024437758002011800091080013030032000080010200800002001600001016000010
160024424208001111800001080000030032000080010200800002001600001016000010
160024404238001111800001080000030032000080010200800002001600001016000010
160024404208001111800001080000030032000080010200800002001600001016000010
160024404178001111800001080000030032000080010200800002001600001016000010
160024404178001111800001080000030032000080010200800002001600001016000010
160024404078001111800001080000030032000080010200800002001600001016000010
160024403908001111800001080000030032000080010200800002001600001016000010
160024404078001111800001080000030032000080010200800002001600001016000010
160024404308001111800001080000030032000080010200800002001600001016000010

Test 5: throughput

Count: 16

Code:

  usra v0.4h, v16.4h, #3
  usra v1.4h, v16.4h, #3
  usra v2.4h, v16.4h, #3
  usra v3.4h, v16.4h, #3
  usra v4.4h, v16.4h, #3
  usra v5.4h, v16.4h, #3
  usra v6.4h, v16.4h, #3
  usra v7.4h, v16.4h, #3
  usra v8.4h, v16.4h, #3
  usra v9.4h, v16.4h, #3
  usra v10.4h, v16.4h, #3
  usra v11.4h, v16.4h, #3
  usra v12.4h, v16.4h, #3
  usra v13.4h, v16.4h, #3
  usra v14.4h, v16.4h, #3
  usra v15.4h, v16.4h, #3
  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)schedule ldst uop (55)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 ldst uop inputs (80)map simd uop inputs (81)? int output thing (e9)? ldst retires (ed)? simd retires (ee)? int retires (ef)
160204802241602021011601010100160105030006400361601082000160012200032002410160000100
160204800351601071011600060100160010030006400361601082000160012200032012810160000100
160204800351601051011600040100160008030006400361601082000160012200032012810160000100
160204800351601051011600040100160008030006402001601522000160060200032012410160000100
160204800571601051011600040100160008030006400361601082000160012200032002410160000100
160204800351601051011600040100160008030006400361601082000160012200032002410160000100
160204800351601051011600040100160008030006400361601082000160012200032002410160000100
160204800351601051011600040100160008030006400361601082000160012200032002410160000100
160204800351601051011600040100160008030006400361601082000160012200032002410160000100
160204800351601051011600040100160008030006400361601082000160012200032002410160000100

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 simd uop inputs (81)? int output thing (e9)? simd retires (ee)? int retires (ef)
160024801151600151116000410160008306400361600182016001220320028116000010
160024800351600171116000610160010306400441600202016001420320028116000010
160024800351600171116000610160010306400441600202016001420320028116000010
160024800351600171116000610160010306400441600202016001420320028116000010
160024800351600171116000610160010306400441600202016001420320028116000010
160024800351600171116000610160010306400441600202016001420320028116000010
160025800711600531116004210160054306402001600622016006020320028116000010
160024800561600171116000610160010306400001600102016000020320098116000010
160024800351600111116000010160000306400001600102016000020320000116000010
160024800351600111116000010160000306400001600102016000020320000116000010