Apple M1 Microarchitecture Research by Dougall Johnson

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

SUQADD (scalar, H)

Test 1: uops

Code:

  suqadd h0, h1
  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)03181e3a3f4e51inst issue (52)~issue fp/simd (54)~dispatch fp/simd (57)huge thing fp/simd (5b)60696d6edispatch 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)a0a1a8a9accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)f5f6f7f8fd
100430372300017725482510001000100039831303018303730372415328951000100020003037303711100110000000073316222630100030383038303830383038
10043037230006125482510001000114939831303018303730372415328951000100020003037303711100110000000073216222630100030383038303830383038
10043037230008225482510001000100039831303018303730372415328951000100020003037303711100110002400094225222630100030383038303830853038
10043037220006125482510001000100039831303018308430372415328951000100020003037303711100110000000073216222630100030383038303830383038
10043037230006125482510001000100039831303018303730372415328951000100020003037303711100110000000073216222630100030383038303830383038
10043037230008225482510001000100039831303018303730372415328951000100020003037303711100110000000073216222630100030383038303830383038
100430372300012425482510001000100039831303018303730372415328951000100020003037303711100110000000073216222630100030383038303830383038
10043037220006125482510001000100039831303018303730372415328951000100020003037303711100110000000073216222630100030383038303830383038
10043037230006125482510001000100039831303018303730372415328951000100020003037303711100110000000073216222630100030383038303830383038
10043037230006125482510001000100039831303018303730372415328951000100020003037303711100110000000073216222630100030383038303830383038

Test 2: Latency 1->1

Code:

  suqadd h0, h1
  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)03181e3f4e51inst 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)9fa8c2c5branch mispredict (cb)cdcfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10204300372250061295482510100100100001001000050042773130300183003730037282726287411010020010008200200163003730037111020110099100100100001000011171701600296460100001003003830038300383003830038
10204300372250061295482510100100100001001000050042773130300183003730037282727287411010020010008200200163003730037111020110099100100100001000011171801600296460100001003003830038300383003830038
10204300372240061295482510100100100001001000050042773130300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011611296340100001003003830038300383003830038
10204300372250061295482510100100100001001000050042773130300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011611296340100001003003830038300383003830038
102043003722500107295482510100100100001001000050042773130300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011601296340100001003003830038300383003830038
10204300372250061295482510100100100001001000050042773131300183003730037282653287451010020010000200200003003730037111020110099100100100001001000071011611296340100001003003830038300383003830038
10204300372250061295482510100100100001001000050042773130300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011611296340100001003003830038300383003830038
10204300372250061295482510100100100001001000050042773130300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011611296340100001003003830038300383003830038
10204300372240061295482510100100100001001000050042773130300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011611296340100001003003830038300383003830038
10204300372240061295482510100100100001001000050042773131300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011611296340100001003003830038300383003830038

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)0308090b18191e1f3f4e51inst 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)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)9fa1a6a8a9acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1002430037225000004888331929494121100601310048131089492428415213023430321302282831604028861109062010821322196230225303198110021109101010000100000840847652162229630010000103003830085300383003830038
10024300372330000030261029494641006910100001010000824285455030378303573041328325020288981120620116452422928304623060512110021109101010000100020006402162229630010000103003830038300383003830038
1002430037225000000061295482510010101000010100005042773131300183003730037282870328767100102010000202000030037300371110021109101010000100000006402162229630010000103003830038300383003830038
1002430037225000000061295482510010101000010100005042773131300183003730037282870328767100102010000202000030037300371110021109101010000100000006402162229630010000103003830038300383003830038
1002430037225000000061295482510010101000010100005042773131300183003730037282870328767100102010000202000030037300371110021109101010000100000006402162229630010000103003830038300383003830038
1002430037225000000061295482510010101000010100005042773131300183003730037282870328767100102010000202000030037300371110021109101010000100000006402162229630010000103003830038300383003830038
1002430037225000000061295482510010101000010100005042773131300183003730037282870328767100102010000202000030037300371110021109101010000100000006402162229630010000103003830038300383003830038
100243003722500000001008295482510010101000010100005042773131300183003730037282870328767100102010000202000030037300371110021109101010000100000006402162229630010000103008430038300383003830038
1002430037225000000061295482510010101000010100005042773131300183003730037282870328767100102010000202000030037300371110021109101010000100010006402162229630010000103003830038300383003830038
1002430037224000000061295482510010101000010100005042773131300183003730037282870328767100102010000202000030037300371110021109101010000100000006402162229630010000103003830038300383003830038

Test 3: Latency 1->2

Code:

  suqadd h0, h0
  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)03070a1e3a3f4e51inst 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)9fa0a8a9acc5branch mispredict (cb)cdcfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10204300372251115061295472510100100100001001000050042771601300183003730037282716287401010020010008200200163003730037111020110099100100100001000101110611171811611296470100001003003830038300383003830038
102043003722511006129547251010010010000100100005004277160130018300373003728271628740101002001000820020016300373003711102011009910010010000100020311171711611296470100001003003830038300383003830038
102043003722511006129547251010010010000100100005004277160130018300373003728271628740101002001000820020016300373003711102011009910010010000100020011171711611296470100001003003830038300383003830038
102043003722511006129547251010010010000100100005004277160130018300373003728271728740101002001000820020016300373003711102011009910010010000100010311171811611296470100001003003830038300383003830038
1020430037225110112029547251010010010000100100005004277160030018300373003728252628733101002001000020020000300373003711102011009910010010000100010311172232433296290100001003003830038300383003830038
102043003722400019729547251010010010000100100005004277160030018300373003728252628733101002001000020020000300373003711102011009910010010000100010011172232433296290100001003003830038300383003830038
102043003722500019729547251010010010000100100005004277160130018300373003728271728741101002001000820020016300373003711102011009910010010000100010311171711611296480100001003003830038300383003830038
102043003722511006129547251010010010000100100005004277160030018300373003728271728741101002001000820020016300373003711102011009910010010000100010311171811611296470100001003003830038300383003830038
102043003722511006129547251010010010000100100005004277160030018300373003728271628741101002001000820020016300373003711102011009910010010000100000611171711611296471100001003003830038300383003830038
102043003722511006129547251010010010000100100005004277160130018300373003728271728741101002001000820020016300373003711102011009910010010000100000911176611611296470100001003003830038300383003830038

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)03080b181e3f4e51inst 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)9fa0a1a6a8a9acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1002430037225000061295472510010101000010100005042771601300183003730037282863287671016020100002020000301313003711100211091010100001000000906408163329629010000103003830038300383003830038
1002430037225000061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001000000006403163329629010000103003830038300383003830038
100243003722500006129547251001010100001010000504277160130018300373003728286328767100102010000202033630037300371110021109101010000100000012906403163329629010000103003830038300383003830038
1002430037224000061295472510010101000010100005042771600300183003730037282863287671001020100002020000300373003711100211091010100001000000006403163329629010000103003830038300383003830038
1002430037225000061295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001000000306403163329629010000103003830038300383003830038
10024300372240000612954725100101010000101000050427716013001830037300372828632876710010201000020200003003730037111002110910101000010000006306403163329629010000103003830038300383003830038
10024300842251012761295472510018101000010100005042771600300543013230037282868288041001020101632020000300373003711100211091010100001000000306403163329629010000103003830038300383003830038
1002430037225000061295472510010101000010100005042771600300183003730037282863287671001020100002020000300373003711100211091010100001000000006403163329629010000103003830038300383003830038
1002430037225000061295472510010101000010100005042771600300183003730037282863287671001020100002020000300373003711100211091010100001000000006403163329629010000103003830038300383003830038
10024300372250001261295472510010101000010100005042771601300183003730037282863287671001020100002020000300373003711100211091010100001000000006403163429629010000103003830038300383003830038

Test 4: throughput

Count: 8

Code:

  movi v0.16b, 0
  suqadd h0, h8
  movi v1.16b, 0
  suqadd h1, h8
  movi v2.16b, 0
  suqadd h2, h8
  movi v3.16b, 0
  suqadd h3, h8
  movi v4.16b, 0
  suqadd h4, h8
  movi v5.16b, 0
  suqadd h5, h8
  movi v6.16b, 0
  suqadd h6, h8
  movi v7.16b, 0
  suqadd h7, h8
  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)030f18191e1f3f51inst 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)6061696d6e74scheduler 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)9fa6a8a9acc5branch mispredict (cb)cdcfd0d2d5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1602042006515020000292580116100800161008002850064019610200452006520065061280128200800282001600562006520065111602011009910010016000010000001111011950016002006201600001002006620066200662006620066
1602042006515000000292580116100800161008002850064019615200452006520065061280128200800282001600562006520065111602011009910010016000010000001111011951016002006201600001002006620066200662006620066
1602042006515000000292580116100800161008002850064019615200452006520065061280128200800282001600562006520065111602011009910010016000010000001111011950016002006201600001002006620066200662006620066
1602042006515000000292580116100800161008002850064019615200452006520065061280128200800282001600562006520065111602011009910010016000010000001111011951016002006201600001002006620066200662006620066
1602042006515100000292580116100800161008002850064019615200452006520065061280128200800282001600562006520065111602011009910010016000010003001111011951016002006201600001002006620066200662006620066
160204200651500000027102580116100800161008002850064019615200452006520065061280128200800282001600562006520065111602011009910010016000010001001111011951016002006201600001002006620066200662006620066
1602042006515000000292580116100800161008002850064019615200452006520065061280128200800282001600562006520065111602011009910010016000010003001111011951016002006201600001002006620066200662006620066
16020420065150000002925801161008001610080028500640196152004520065200650612801282008002820016005620065200651116020110099100100160000100030014781111011951016002006201600001002006620066200662006620066
1602042006515100000292580116100800161008002850064019615200452006520065061280128200800282001600562006520065111602011009910010016000010002001111011951016002006201600001002006620066200662006620066
1602042006515100000292580116100800161008002850064019615200452006520065061280128200800282001600562006520065111602011009910010016000010004001111011951016002006201600001002006620066200662006620066

1000 unrolls and 10 iterations

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

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)5f6061696d6escheduler 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)9fa8a9acc2cfd0d2icache miss (d3)d5d6d9dadbddinst fetch restart (de)e0eaec? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1600242008815004529800101080000108000050640000115200322005120051322800102080000201600002006020060111600211091010160000100000100373112825221151220057201160000102005220052200612005220052
16002420051150071029800101080000108000050640000110200322005120051322800102080000201600002005120051111600211091010160000100400100373111425211221720048201160000102005220052200522005220052
1600242005115004530800101080000108000050640000015200322005120051322800102080000201600002005120051111600211091010160000101030100378411625411141420048201160000102005220052200522005220052
16002420051150045278001010800001080000506400002152003220051200513228001020800002016000020051200511116002110910101600001000870100448412125211111320048201160000102005220052200522005220142
1600242005115004527800101080000108000050640000215200322005120051322800102080000201600002005120051111600211091010160000100000100378511925211191420048201160000102005220052200522005220052
1600242005115004527800101080000108000050640000115200322005120051322800102080000201600002005120051111600211091010160000100000100368411325221181220048201160000102005220052200522005220052
16002420051150045278001010800001080000506400001152003220051200513228001020800002016000020051200511116002110910101600001016000100348411125211131120048401160000102005220052201422005220052
1600242005115104527800101080000108000050640000115200322005120051322800102080000201600002005120051111600211091010160000100000100368411025212171720048401160000102005220052200522005220061
1600242005115004527800101080000108000050640000115200412005120051322800102080000201600002005120051111600211091010160000100000100378411325411141420057401160000102005220052200522005220052
1600242005115004527800101080000108000050640000115200322005120051322800102080000201600002005120051111600211091010160000101000100368411325212111320048201160000102005220052200612005220052

Test 5: throughput

Count: 16

Code:

  suqadd h0, h16
  suqadd h1, h16
  suqadd h2, h16
  suqadd h3, h16
  suqadd h4, h16
  suqadd h5, h16
  suqadd h6, h16
  suqadd h7, h16
  suqadd h8, h16
  suqadd h9, h16
  suqadd h10, h16
  suqadd h11, h16
  suqadd h12, h16
  suqadd h13, h16
  suqadd h14, h16
  suqadd h15, h16
  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)03191e1f3f51inst 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)9fa1a8acc5branch mispredict (cb)cdcfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1602044006730004238830251601081001600081001600205001280132140020400394003919977619990160120200160032200320064400394003911160201100991001001600001000001111011811600400361600001004004040040400404004040040
1602044003930000030251601081001600081001600205001280132140020400394003919977619990160120200160032200320064400394003911160201100991001001600001000001111011801600400361600001004004040040400404004040040
1602044003930000030251601081001600081001600205001280132140020400394003919977619990160120200160032200320064400394003911160201100991001001600001000001111011801600400361600001004004040040400404004040040
1602044003930000030251601081001600081001600205001280132040020400394003919977619990160120200160032200320064400394003911160201100991001001600001000001111011801600400361600001004004040040400404004040040
1602044003929900030251601081001600081001600205001280132040020400394003919977619990160120200160032200320064400394003911160201100991001001600001002001111011801601400361600001004004040040400404004040040
1602044003930000030251601081001600081001600205001280132040020400394003919977619990160120200160032200320064400394003911160201100991001001600001000001111011801600400361600001004004040040400404004040040
1602044003930006030251601081001600081001600205001280132140020400394003919977619990160120200160032200320064400394003911160201100991001001600001000001111011801600400361600001004004040040400404004040040
1602044003930000072251601081001600081001600205001280132140020400394003919977619990160120200160032200320064400394003911160201100991001001600001000001111011801600400361600001004004040040400404004040040
1602044003929900030251601081001600081001600205001280132140020400394003919977619990160120200160032200320064400394003911160201100991001001600001000001111011801600400361600001004004040040400404004040040
16020440039300000996251601081001600081001600205001280132040020400394003919977619990160120200160032200320064400394003911160201100991001001600001000001111011801600400361600001004004040040400404004040040

1000 unrolls and 10 iterations

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

retire (01)cycle (02)0307080a0b18191e1f3a3f51inst 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)5f6061696b6d6edispatch 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)9fa1a6a7a8a9acc2cfd0d2icache miss (d3)d5d6d9dadbddinst fetch restart (de)e0? int output thing (e9)eaeb? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
16002440039299000000000462516001010160000101600005012800001154002004003940039199963200191600102016000020320000400394003911160021109101016000010000000010024114216163212611400360206160000104004340040400404004040040
160024400392991010122882081431671602051016019510160205501281610015401160401514015120015112007116022020160210203204044015940149311600211091010160000100010402100601162213732231134008404314160000104021440150400404009440040
16002440039300101000120167625160010101600001016000050128000001540020040039400391999632001916001020160000203200004003940039111600211091010160000100002000100261162141632227164003604113160000104005040040400404004040040
160024400393000000000016425160010101600001016000050128000001540020040039400391999632001916001020160000203200004003940039111600211091010160000100000030100261162281632227284003604113160000104005040104400404004040040
160024400393001010000016425160010101600001016000050128000001540020040039400391999632001916001020160000203200004003940039111600211091010160000100000000100241162281632226264003604113160000104005640040400404004040040
160024400393001010000005225160010101600001016000050128000001540020040039400392001132001916001020160000203200004003940039111600211091010160000100000000100261162291632228154003604113160000104005040040400404004040040
160024400393001010000015225160010101600001016000050128000001540020040039400391999632001916001020160000203200004003940039111600211091010160000100000000100261162281632227284003604113160000104005040040400404004040040
160024400392991010000006425160010101600001016000050128000001540020040039400391999632001916001020160000203200004003940039111600211091010160000100000000100261672281632228284003604113160000104005040040400404004040040
1600244003930010100000164251600101016000010160000501280000011040020040039400391999632001916001020160000203200004003940039111600211091010160000100001000100261672281632219284003604113160000104004040040400404004040040
1600244003930000000000152251600101016000010160000501280000011040020040039400391999632001916001020160000203200004003940039111600211091010160000100000000100241672161632228184003604113160000104005140040400404004040040