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

PRFM (register, PSTL2STRM)

Test 1: uops

Code:

  prfm pstl2strm, [x6]
  mov x0, 0

(no loop instructions)

1000 unrolls and 1 iteration

Retires: 1.000

Issues: 1.000

Integer unit issues: 0.001

Load/store unit issues: 1.000

SIMD/FP unit issues: 0.000

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule ldst uop (55)dispatch ldst uop (58)simd uops in schedulers (5a)dispatch uop (78)map ldst uop (7d)map ldst uop inputs (80)? int output thing (e9)? ldst retires (ed)
1004208610011100010003474210001000100011000
1004210510011100010483565910481057100011000
1004209710011100010003491810001000100011000
1004210110011100010003489410001000100011000
1004211110011100010003485810001000100011000
1004208410011100010003474010001000100011000
1004207410011100010003500010001000100011000
1004209910011100010003489610001000100011000
1004209910011100010003494010001000100011000
1004207410011100010003496010001000100011000

Test 2: throughput

Code:

  prfm pstl2strm, [x6]
  add x6, x6, 64

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 2.0120

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule ldst uop (55)dispatch int uop (56)dispatch ldst uop (58)int uops in schedulers (59)simd uops in schedulers (5a)dispatch uop (78)map int uop (7c)map ldst uop (7d)map int uop inputs (7f)map ldst uop inputs (80)? int output thing (e9)? ldst retires (ed)? int retires (ef)
20204207782010110101100001010210000613703492152010110203100031020310003100011000010100
20204200492010110101100001010210000613233492152010210204100041020410004100011000010100
20204200492010110101100001010210000613233492152010210204100041020410004100011000010100
20204200492010110101100001010210000613233492152010210204100041020410004100011000010100
20204200492010110101100001010210000613233492152010210204100041020410004100011000010100
20204200492010110101100001010210000613233492152010210204100041020410004100011000010100
20204200492010110101100001010210000613233492152010210204100041023410034100321000010100
20204200492010110101100001010210004612033503632011410212100121020410004100011000010100
20204200782010110101100001010210000613233492152010210204100041020410004100011000010100
20204200492010110101100001010210000613233492152010210204100041020410004100011000010100

1000 unrolls and 10 iterations

Result (median cycles for code): 2.0122

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule ldst uop (55)dispatch int uop (56)dispatch ldst uop (58)int uops in schedulers (59)simd uops in schedulers (5a)dispatch uop (78)map int uop (7c)map ldst uop (7d)map int uop inputs (7f)map ldst uop inputs (80)? int output thing (e9)? ldst retires (ed)? int retires (ef)
20024208492001410014100001001310000610563505752001410026100061002010000100011000010010
20024201222001110011100001001010000610323505752001010020100001002010000100011000010010
20024201222001110011100001001010000610323505752001010020100001002010000100011000010010
20024201222001110011100001001010000610323505752001010020100001002010000100011000010010
20024201222001110011100001001010000610323505752001010020100001002010000100011000010010
20024201222001110011100001001010000610323505752001010020100001002010000100011000010010
20024201222001110011100001001010000610323505752001010020100001002010000100011000010010
20024201222001110011100001001010000610323505752001010020100001002010000100011000010010
20024201222001110011100001001010000610323505752001010020100001002010000100011000010010
20024201222001110011100001001010000610323505752001010020100001002010000100011000010010

Test 3: throughput

Code:

  prfm pstl2strm, [x6]
  mov x7, 8

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 1.8718

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule ldst uop (55)dispatch int uop (56)dispatch ldst uop (58)int uops in schedulers (59)simd uops in schedulers (5a)dispatch uop (78)map int uop (7c)map ldst uop (7d)map int uop inputs (7f)map ldst uop inputs (80)? int output thing (e9)? ldst retires (ed)? int retires (ef)
1020419154101011011000010010000300332006101002001001020010012110000100
1020418718101011011000010010006300325270101062001001220010012110000100
1020418718101011011000010010006300325270101062001001220010012110000100
1020418718101011011000010010006300325270101062001001220010012110000100
1020418946101011011000010010006300325270101062001001220010012110000100
1020419291101011011000010010000300335988101002001000820010004110000100
1020419376101011011000010010002300338922101022001001220010008110000100
1020418998101011011000010010004300337398101042001001220010004110000100
1020419186101011011000010010004300335434101042001001220010008110000100
1020419234101011011000010010004300335466101042001001220010012110000100

1000 unrolls and 10 iterations

Result (median cycles for code): 2.0048

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)schedule ldst uop (55)dispatch int uop (56)dispatch ldst uop (58)int uops in schedulers (59)simd uops in schedulers (5a)dispatch uop (78)map int uop (7c)map ldst uop (7d)map int uop inputs (7f)map ldst uop inputs (80)? int output thing (e9)? ldst retires (ed)? int retires (ef)
1002420039100111110000101000630349222100162010012201000011000010
1002420048100111110000101000030349214100102010000201000011000010
1002420048100111110000101000030349214100102010000201000011000010
1002420048100111110000101000030349214100102010000201000011000010
1002420074100111110000101000030349292100102010000201000011000010
1002420090100111110000101000030349618100102010000201000011000010
1002420050100111110000101000030349972100102010000201000011000010
1002420048100111110000101000030349434100102010000201000011000010
1002420048100111110000101000030349214100102010000201000011000010
1002420048100111110000101000030349214100102010000201000011000010