Apple M1 Microarchitecture Research by Dougall Johnson

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

SQDMULH (by element, S)

Test 1: uops

Code:

  sqdmulh s0, s0, v1.s[1]
  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)031e3f4e51inst 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)a8accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)f5f6f7f8fd
100430372206125482510001000100039831303018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
100430372306125482510001000100039831303018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
100430372206125482510001000100039831303018303730372415328951000100020003037303711100110003073116112630100030383038303830383038
100430372306125482510001000100039831303018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
100430372308225482510001000100039831303018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
100430372306125482510001000100039831303018303730372418328951000100020003037303711100110000073116112630100030383038303830383038
100430372206125482510001000100039831303018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
100430372306125482510001000100039831303018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
100430372206125482510001000100039831303018303730372415328951000100020003037303711100110000073116112630100030383038303830383038
100430372306125482510001000100039831303018303730372415328951000100020003037303711100110000073116112630100030383038303830383038

Test 2: Latency 1->2

Code:

  sqdmulh s0, s0, v1.s[1]
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)030918191e1f3f4e51inst 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)9fa0a6a8acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10204300372250011440612954825101001261000011410149572428002713001830037300372826532876310430200103342042033430085301311110201100991001001000010002032852149112977816100001003018230275302293022630279
102043008422713526744061295482510100100100001001000050042773131300183003730037282683287451010020010000200200003003730037111020110099100100100001000002770071011611296340100001003008530038300383003830038
1020430037225000144061295482510100100100001001000050042773131300183003730037282653287451010020010000200200003003730037111020110099100100100001000003071011611296340100001003003830038300383003830038
1020430037225000132061295482510100100100001001000063242773130300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011611296340100001003003830038300383003830038
10204300372250000061295482510100107100001001000050042773131300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011611296340100001003003830038300383003830038
10204300372250001320822954825101001001000010010000500427731313001830037300372826532874510100200100002002000030037300371110201100991001001000010000000710116112963425100001003003830038300383003830038
102043003722500008861295482510100100100001001000050042773131300183003730037282653287451010020010000200200003003730037111020110099100100100001000010071011611296340100001003003830038300383003830038
10204300372250000061295482510100100100001001000050042773131300183003730037282653287451010020010000200200003003730037111020110099100100100001002000071011611296340100001003003830086300863003830038
102043042122700090749295482510100100100001001000050042773131300183003730037282657287801026720010000200200003003730131111020110099100100100001000000071011611296340100001003003830038300383003830038
10204300372250000061295482510100100100001001000050042773130300183003730037282653287451010020010000200200003003730037111020110099100100100001000000071011611296340100001003003830038300383003830038

1000 unrolls and 10 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)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)9fa8accdcfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10024300372250001102954825100101010000101000050427731313001830037300372828732876710010201000020200003003730037111002110910101000010000640216322963010000103003830038300383003830038
1002430037225000612954825100101010000101000050427731313001830037300372828732876710010201000020200003003730037111002110910101000010000640216222963010000103003830038300383003830038
1002430037225000612954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640216322963010000103003830038300383003830038
1002430037225000612954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640216222963010000103003830038300383003830038
10024300372250001242954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640216322963010000103003830086300383003830038
1002430037225000612954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000640216222963010000103003830038300383003830038
1002430037225000612954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010030640216322963010000103003830038300383003830038
1002430037225000612954825100101010000101000050427731313001830037300372828732876710010201000020200003003730037111002110910101000010000640216222963010000103003830038300383003830038
1002430037225000612954825100101010000101000050427731313001830037300372828732876710160201000020200003003730037111002110910101000010000640216322963010000103003830038300383003830038
1002430037225000612954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000668216422963010000103003830038300383003830038

Test 3: Latency 1->3

Code:

  sqdmulh s0, s1, v0.s[1]
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)03080b1e3f4e51inst 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)9fa0a8accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001001007101161129634100001003003830038300383003830038
102043003722500126129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037211020110099100100100001000037101161129634100001003003830038300383003830038
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830123
10204300372260106129548251011810010008100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000107101161129634100001003003830038300383003830038
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000067101161129634100001003003830038300383003830038
10204300372250006129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038
10204300372250066129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000507101161129634100001003003830038300383003830038
102043003722500013129548251010010010000100100005004277313300183003730037282653287451010020010000200200003003730037111020110099100100100001000007101161129634100001003003830038300383003830038

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)030b0f18191e1f3a3f4e51inst 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)9fa6a8accfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10024300372250100000612954825100101010000101000050427731330018300373003728287328767100102010000202000030037300371110021109101010000100006402162229630010000103003830038300383003830038
10024300372250000000612954825100101010000101000050427731330018300373003728287328767100102010000202000030037300371110021109101010000100006402162229630010000103003830086300863003830038
10024300372250000000612954825100101010000101000050427731330018300373003728287328767100102010000202000030037300371110021109101010000100006402162229700010000103003830038300383003830038
10024300372250000000612954825100101010000101000050427731330018300373003728287328767100102010000202000030037300371110021109101010000100006402162229630010000103003830038300383003830038
10024300372240000000612954825100101010000101000050427731330018300373003728287328767100102010000202000030084300371110021109101010000100006402162229630010000103003830038300383003830038
10024300372250000000612954825100101010000101000050427731330018300373003728287328767100102010000202000030037300371110021109101010000100006402162229630010000103003830038300383003830038
10024300372240000000612954825100101010000101000050427731330018300373003728287328767100102010000202000030037300371110021109101010000100006402162229630010000103003830038300383003830038
10024300372250000000612954825100101010000101000050427731330018300373003728287328767100102010000202000030037300371110021109101010000100636402162229630010000103003830038300383003830038
10024300372250000000612954825100101010000101000050427731330018300373003728287328767100102010000202000030037300371110021109101010000100006402162229630010000103003830038300383003830038
10024300372250000000612954825100101010000101000050427731330018300373003728287328767100102010000202000030037300371110021109101010000100006402162229630010000103003830038300383003830038

Test 4: throughput

Count: 8

Code:

  sqdmulh s0, s8, v9.s[1]
  sqdmulh s1, s8, v9.s[1]
  sqdmulh s2, s8, v9.s[1]
  sqdmulh s3, s8, v9.s[1]
  sqdmulh s4, s8, v9.s[1]
  sqdmulh s5, s8, v9.s[1]
  sqdmulh s6, s8, v9.s[1]
  sqdmulh s7, s8, v9.s[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): 0.2505

retire (01)cycle (02)03080b18191e1f3a3f51inst 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)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)9fa0a1a6a8a9acc2cfd0d5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
802042005815000001800412580100100800001008000050064000000200202003920039997339997801002008000020016000020039200391180201100991001008000010000000005110021611200360800001002004020040200402004020040
80204200391500000186007062580100100800001008000050064000010200202003920039997339997801002008000020016000020039200391180201100991001008000010000000005110511611200360800001002004020040200402004020040
80204200391500000000412580100100800001008000050064000005200202003920039997339997801002008000020016000020039200391180201100991001008000010000000005110011611200360800001002004020040200402004020040
80204200391500000000412580100100800001008000050064000005200202003920039997339997801002008000020016000020039200391180201100991001008000010000000005110011611200360800001002004020040200402004020040
8020420039150000045000412580100100800001008000050064000005200202003920039997339997801002008000020016000020039200391180201100991001008000010000000005110511611200360800001002004020170200902010120040
802042003915000003900412580100100800001008000050064000005200202003920039997339997801002008000020016000020039200391180201100991001008000010000000005110011611200360800001002004020040200402004020040
802042003915000006300412580100100800001008000050064000010200202003920039997339997801002008000020016000020039200391180201100991001008000010000000005110011611200360800001002004020040200402004020040
80204200391500000000412580100100800001008000050064000010200202003920039997339997801002008000020016000020039200391180201100991001008000010000000005110511611200360800001002004020040200402004020040
80204200391500000000412580100100800001008000050064000010200202003920039997339997801002008000020016000020039200391180201100991001008000010000000005110011611200360800001002004020040200402004020040
80204200391500000000412580100100800001008000050064000005200202003920039997339997801002008041820016000020039200391180201100991001008000010000000005110011611200360800001002004020040200402004020040

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.2505

retire (01)cycle (02)03080a0b191e1f3a3f51inst 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)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)9fa0a1a6a8acc2branch mispredict (cb)cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
800242003915000001110040258001010800001080000506400000020020200392003999960310019800102080000201600002003920039118002110910108000010000000050202161120036080000102004020040200402004020040
8002420039150000000040258001010800001080000506400000020020200392003999960310019800102080000201600002003920039118002110910108000010000000050201161120036080000102004020040200402004020040
80024200391500000390040258001010800001080000506400000120020200392003999960310019800102080000201600002003920039118002110910108000010000000050201161120036080000102004020040200402004020040
80024200391500000510040258001010800001080000506400000120020200392003999960310019802252080000201600002003920039118002110910108000010000000050201161120036080000102004020040200402004020040
8002420039150000000040258001210800001080000506400000020020200392003999960310019800102080000201600002003920039118002110910108000010000000050221161120036080000102004020040200402004020040
8002420039150000000040258001010800001080000506400000120020200392003999960310019800102080000201600002003920039118002110910108000010000000050201161120036080000102004020040200402004020040
8002420039150000000040258001010800001080000506400000120020200392003999960310019800102080000201600002003920039118002110910108000010000000050201161120036080000102004020040200402004020040
8002420039150000000040258001010800001080000506400000120020200392003999960310019800102080000201600002003920039118002110910108000010000000050201161120036080000102004020040200402004020040
800242003915000004500657258001010800001080000506400000120020200392003999960310019800102080000201600002003920039118002110910108000010000000050201161120036080000102004020040200402004020040
800242003915000001530040258001012800001080000606400000120020200392003999960310019800102080000201600002003920039118002110910108000010000000050201161120036080000102004020040200402004020040