Apple M1 Microarchitecture Research by Dougall Johnson

Firestorm: Overview | Base Instructions | SIMD and FP Instructions
Icestorm:  Overview | Base Instructions | SIMD and FP Instructions

SRSRA (vector, 2D)

Test 1: uops

Code:

  srsra v0.2d, v1.2d, #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: 0.000

Load/store unit issues: 0.000

SIMD/FP unit issues: 1.000

retire (01)cycle (02)03073f4e51inst issue (52)~issue fp/simd (54)~dispatch fp/simd (57)huge thing fp/simd (5b)696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op fp/simd (7e)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst neon or fp (9a)a8accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)f5f6f7f8fd
10043037230822548251000100010003983133018303730372415328951000100020003037303711100110000075116112630100030383038303830383038
10043037230612548251000100010003983133018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
10043037230612548251000100010003983133018303730372415328951000100020003037303711100110000973116112630100030383038303830383038
10043037231612548251000100010003983133018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
10043037230612548251000100010003983133018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
10043037230612548251000100010003983133018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
10043037220612548251000100010003983133018303730372415328951000100020003037303711100110000073116222630100030383038303830383038
100430372201032548251000100010003983133018303730372415328951000100020003037303711100110000075216222627100030383038303830383038
10043037220612548251000100010003983133018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
10043037230612548251000100010003983133018303730372415328951000100020003037303711100110000073116112630100030383038303830383038

Test 2: Latency 1->1

Code:

  srsra v0.2d, v1.2d, #3
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)0308181e3f4e51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9fa8acc2cfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038
10204300372251006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001006007101161129634100001003003830038300383003830038
1020430037225001866129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000557807101161129634100001003003830038300383003830038
10204300372250006129548251015110010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038
10204300372240006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001002007101161129634100001003003830038300383003830038
10204300372240006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)0308090b191e1f3f4e51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9fa1a8accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1002430037225000000612954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640316332963010000103008630086300383003830038
100243003722500000010362954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640316332963010000103003830038300383003830038
1002430037225000000612954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640316332963010000103003830038300383003830038
1002430037226000000612954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640316332963010000103003830038300383003830038
10024300372250000003602954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640316332963010000103003830038300383003830038
100243003722500000020512954825100101010000101000050427731303001830085300372828732876710160201000020200003003730037111002110910101000010000640316332963010000103003830038300383003830038
10024300372250000007262954825100101010032101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640316332963010000103003830038300383003830038
10024300372250000002512954825100101010000101000050427731303001830037300372829932876710010201000020200003003730037111002110910101000010000640316332963010000103003830038300383003830038
1002430037225000000612954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640316332963010000103003830038300383003830038
10024300372250000006129548251001010100001010000504277313030018300373003728287172876710010201000020200003003730037111002110910101000010000640316332963010000103003830038300383003830038

Test 3: Latency 1->2

Code:

  srsra v0.2d, v0.2d, #3
  movi v0.16b, 1

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)03080b18191e1f3a3f4e51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9fa1a6a7a8a9acc2c5branch mispredict (cb)cdcfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
102043003722500000006129547251010010010000100100005004277160030018300373003728271628741101002001000020020000300373003711102011009910010010000100000000011172242434296290100001003003830038300383003830038
102043003722500000009729547251010010010000100100005004277160030018300373003728252628733101002001000020020000300373003711102011009910010010000100000000011172242434296290100001003003830038300383003830038
102043003722500000009729547251010010010000100100005004277160030018300373003728252628733101002001000020020000300373003711102011009910010010000100000000011172242443296290100001003003830038300383003830038
102043003722500000019729547251010010010000100100005004277160030018300373003728252628733101002001000020020000300373003711102011009910010010000100000000011172232444296290100001003003830038300383003830038
1020430084225000000197295472510100100100001001000050042771600300183003730037282526287331010020010000200200003003730084111020110099100100100001000010000111722580122982810100001003037130229303243037030324
10204303232250176936616010329547251010010010000100100005004277160030018300373003728271628741101002001000820421012301793013261102011009910010010000100220108383211179056446297386100001003018230277302273018130226
1020430178226102614188053729484229102091241005613411050648428695703027030653304692828959289421125624411821236243003056530549111102011009910010010000100201023617521119722120762994014100001003046530515306543037330421
10204306542371099118896814684294661761016311910064118109005934285272030306304173051028314442887910869220113352262331430466304191011020110099100100100001000200232111829449112997221100001003051430454306093069430656
1020430752237006131455132033942294841591024313010056133110505934292623130018300373003728271628740101002001000820020016300373003711102011009910010010000100000000011171711634296291100001003003830038300383003830038
102043003722500000006129547251010010010000100100005004277160130018300373003728271628740101002001000820020016300373003711102011009910010010000100000000011171711611296460100001003003830038300383003830038

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)03191e3a3f4e51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9fa8cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
100243003722500061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001006402162229629010000103003830038300383003830038
100243003722500061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001006402162229629010000103003830038300383003830038
100243003722500061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001006402162229629010000103003830038300383003830038
100243003722503061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001006402162229629010000103003830038300383003830038
1002430037225000375295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001006402162229667010000103003830038300863008530086
1002430085225048161295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001016402162229629010000103003830038300383003830038
1002430037225012061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001006402162229629010000103003830038300383003830038
100243003722500061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001036402162229629010000103003830038300383003830038
100243003722500061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001006402163229629010000103003830038300383003830038
1002430037225121061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001006402162229629010000103003830038300383003830038

Test 4: throughput

Count: 8

Code:

  movi v0.16b, 0
  srsra v0.2d, v8.2d, #3
  movi v1.16b, 0
  srsra v1.2d, v8.2d, #3
  movi v2.16b, 0
  srsra v2.2d, v8.2d, #3
  movi v3.16b, 0
  srsra v3.2d, v8.2d, #3
  movi v4.16b, 0
  srsra v4.2d, v8.2d, #3
  movi v5.16b, 0
  srsra v5.2d, v8.2d, #3
  movi v6.16b, 0
  srsra v6.2d, v8.2d, #3
  movi v7.16b, 0
  srsra v7.2d, v8.2d, #3
  movi v8.16b, 9

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.2508

retire (01)cycle (02)031e3f51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)696b6d6escheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9fa0a8c5branch mispredict (cb)cdcfd0d5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1602042007715002925801161008001610080028500640196200450200652006561280128200800282001600562006520311111602011009910010016000010000111101270816882006211600001002006620066200662006620066
1602042006515502925801161008001610080028500640196200450200652006561280128200800282001600562006520065111602011009910010016000010000111101270816882006201600001002006620066200662006620066
1602042006515002925801161008001610080028500640196200450200652006561280128200800282001600562006520065111602011009910010016000010000111101260916882006201600001002006620066200662006620066
1602042006515102925801161008001610080028500640196200450200652006561280128200800282001600562006520065111602011009910010016000010000111101270716882006201600001002006620066200662006620066
1602042006515002925801161008001611480028500640196200450200652006561280128200800282001600562006520065111602011009910010016000010000111101280816882006201600001002006620066200662006620066
1602042006515002925801161008001610080028500640196200450200652006561280128200800282001600562006520065111602011009910010016000010000111101270316892006201600001002013420066200662006620066
16020420065150452925801161008001610080028500640196200450200652006561280128200800282001600562006520065111602011009910010016000010000111101270816882006201600001002006620066200662006620066
16020420065150067525801161008001610080028500640196200450200652006561280128200800282001600562006520065111602011009910010016000010000111101270316872006201600001002006620066200662006620066
1602042006515002925801161008001610080028500640196200450200652006561280128200800282001600562006520065111602011009910010016000010000111101270816882006201600001002006620066200662006620066
1602042006515102925801161008001610080028500640196200450200652006561280128200800282001600562006520065111602011009910010016000010000111101270816732006201600001002006620066200662006620066

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.2506

retire (01)cycle (02)03081e3f51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)5f60696d6escheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9faccfd0d2icache miss (d3)itlb miss (d4)d5d6d9dadbddinst fetch restart (de)e0ea? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1600242007215000452580010108000010800005064000011200272004620046322800102080000201600002004620046111600211091010160000100100273210820211662004315160000102004720047200472004720047
1600242004615000452580010108000010800005064000001200272005020050322800102080000201600002005020050111600211091010160000100100326220624422642004730160000102005120051200512005120051
1600242005015000512580010108000010800005064089201200312005020050322800102080000201600002005020050111600211091010160000100100316220624422662004730160000102005120051200512005120051
16002420050150001622580010108000010800005064000001200312005020050322800102080000201600002005020050111600211091010160000100100326220624422742004730160000102005120051200512005120051
1600242005015000512580010108000010800005064000001200312005020050322800102080000201600002005020050111600211091010160000100100306220624422662004730160000102005120051200512005120051
16002420050150001162580010108000010800005064000001200312005020050322800102080000201600002005020050111600211091010160000100100326220424422462004730160000102005120051200512005120051
1600242005015000512580010108000010800005064000001200312005020050322800102080000201600002005020050111600211091010160000100100326220624422462004730160000102005120051200512005120051
1600242005015000512580010108000010800005064000001200312005020050322800102080000201600002005020050111600211091010160000100100326220624422462004730160000102005120051200512005120051
1600242005015000512580010108000010800005064000001200312005020050322800102080000201600002005020050111600211091010160000100100326220624422662004730160000102005120051200512005120051
1600242005015000512580010108000010800005064000001200312005020050322800102080000201600002005020050111600211091010160000100100326220624422662004730160000102005120051200512005120051

Test 5: throughput

Count: 16

Code:

  srsra v0.2d, v16.2d, #3
  srsra v1.2d, v16.2d, #3
  srsra v2.2d, v16.2d, #3
  srsra v3.2d, v16.2d, #3
  srsra v4.2d, v16.2d, #3
  srsra v5.2d, v16.2d, #3
  srsra v6.2d, v16.2d, #3
  srsra v7.2d, v16.2d, #3
  srsra v8.2d, v16.2d, #3
  srsra v9.2d, v16.2d, #3
  srsra v10.2d, v16.2d, #3
  srsra v11.2d, v16.2d, #3
  srsra v12.2d, v16.2d, #3
  srsra v13.2d, v16.2d, #3
  srsra v14.2d, v16.2d, #3
  srsra v15.2d, v16.2d, #3
  movi v16.16b, 17

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.2502

retire (01)cycle (02)03080a0b181e3f51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)6061696d6edispatch stall (70)72scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9fa6a8acc5branch mispredict (cb)cdcfd0d2d5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
16020440059300000003252516010810016000810016002050012801320540020400394003919977061999016012020016003220032006440039400391116020110099100100160000100000111101185101611400361600001004004040040400404004040040
1602044003930000000302516010810016000810016002050012801320040020400394003919977761999016012020016003220032006440039400391116020110099100100160000100000111101185111610400361600001004004040040400404004040040
1602044003930001000302516010810016000810016013050012801321540020400394003919977061999016012020016003220032006440039400391116020110099100100160000100000111101180001600400361600001004004040040400404004040040
1602044003930000000302516010810016000810016002050012801320540020400394003919977061999016012020016003220032006440039400391116020110099100100160000100000111101185011610400361600001004004040040400404004040040
1602044003929900000302516010810016000810016002050012801321040020400394003919977061999016012020016003220032006440039400391116020110099100100160000100000111101180011611400361600001004004040040400404004040040
1602044003929900000302516010810016000810016002050012801320040020400394003919977061999016012020016003220032006440039400391116020110099100100160000100000111101180001600400361600001004004040040400404004040040
16020440039300000006952516010810016000810016002050012801321540020400394003919977061999016012020016003220032006440039400391116020110099100100160000100000111101180011610400361600001004004040040400404004040040
16020440039300000054302516010810016000810016002050012801321040020400394003919977061999016012020016003220032006440039400391116020110099100100160000100000111101180101610400361600001004004040040400404004040040
1602044003930000000302516010810016000810016002050012801321540020400394003919977061999016012020016003220032006440039400391116020110099100100160000100000111101185001611400361600001004004040040400404004040040
1602044003929900000302516010810016000810016002050012801320540020400394003919977061999016012020016003220032006440039400391116020110099100100160000100000111101185101601400361600001004004040040400404004040040

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.2502

retire (01)cycle (02)030b18191e1f3a3f51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)5f60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9fa0a6a7a8accfd0d2icache miss (d3)d5d6d9dadbddinst fetch restart (de)e0eaeb? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1600244005129900000046251600101016000010160000501280000114002040039400391999632001916001020160000203200004003940039111600211091010160000100000010022622211622296400363010160000104004040040400404004040040
16002440039300000000522516001010160000101600005012800000140020400394003919996320019160010201600002032000040039400391116002110910101600001000000100246221016422106400363010160000104004040040400404004040040
160024400393000000004625160010101600001016000050128000011400204003940039199963200191600102016000020320000400394003911160021109101016000010000001002231191621191040036155160000104004040040400404004040040
160024400393000000005225160010101600001016000050128000001400204003940039199963200191600102016000020320000400394003911160021109101016000010000001002431110164221411400363010160000104004040040400404004040040
160024400392990000004625160010101600001016000050128075611400204003940039199963200191600102016000020320000400394003911160021109101016000010000001002432110162115740036155160000104004040040400404004040040
1600244003930000000046251600101016000010160000501280000104002040039400391999632001916001020160000203200004003940039111600211091010160000100000010024622916422145400363010160000104004040040400404004040040
160024400393000000005225160010101600001016000050128000001400614003940039199963200191600102016000020320000400394003911160021109101016000010000001002462111162128440036155160000104004040040400404004040040
1600244003930000000046251600101016000010160000501280000114002040039400391999632001916001020160000203200004003940039111600211091010160000100000010024622916422106400363010160000104004040040400404009040040
160024400393000000005225160010101600001016000050128000001400204003940039199963200191600102016000020320000400394003911160021109101016000010000001002231171621151240036155160000104004040040400404004040040
16002440039300000000236251600101016000010160000501280000014002040039400391999632001916001020160000203200004003940039111600211091010160000100000010024622716422512400363010160000104004040040400404004040040