Apple M1 Microarchitecture Research by Dougall Johnson

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

FMULX (vector, 8H)

Test 1: uops

Code:

  fmulx v0.8h, v0.8h, v1.8h
  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)03081e3f4e51inst 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
1004403731018613407251000100010005319084018403740373258338951000100020004037403711100110000073116113473100040384038403840384038
100440373000613407251000100010005319084018403740373258338951000100020004037403711100110000073116113473100040384038403840384038
100440373000613407251000100010005319084018403740373258338951000100020004037403711100110001073116113473100040384038403840384038
100440373000613407251000100010005319084018403740373258338951000100020004037403711100110001373116113473100040384038403840384038
100440373000613407251000100010005319084018403740373258338951000100020004037403711100110000673116113473100040384038403840384038
100440373100613407251000100010005319084018403740373258338951000100020004037403711100110000073116113473100040384038403840384038
100440373000613407251000100010005319084018403740373258338951000100020004037403711100110000073116113473100040384038403840384038
100440373000613407251000100010005319084018403740373258338951000100020004037403711100110000073116113473100040384038403840384038
1004403730004073407251000100010005319084018403740373258338951000100020004037403711100110000073116113473100040384038403840384038
100440373000613407251000100010005319084018403740373258338951000100020004037403711100110000073116213473100040384038403840384038

Test 2: Latency 1->2

Code:

  fmulx v0.8h, v0.8h, v1.8h
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 4.0037

retire (01)cycle (02)030918191e3f4e5051inst 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)9fa0a1a7a8accdcfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
102044003730000006139407025101001001000010010000500570690840018400374003738108338745101002001000020020000400374003711102011009910010010000100000106071011611394790100001004003840038400384003840038
10204400373000000613940702510100100100001001000050057069084001840037400373810833874510100200100002002000040037400371110201100991001001000010000099071011611394790100001004003840038400384003840038
10204400373000000613940742510100100100001001000050057069084001840037400373810833874510100200100002002000040037400371110201100991001001000010000020071011611394790100001004003840038400384003840038
10204400373000000613940702510100100100001001000050057069084001840037400373810833874510100200100002002000040037400371110201100991001001000010000089071011611394790100001004003840038400384003840038
102044003730000006139407025101001001000012910000500570690840018400374003738108338745101002001000020020000400374003711102011009910010010000100001333071011611394790100001004003840038400384003840038
102044003730000006139407025101001001000010010000500570690840018400374003738108338745101002001000020020000400374003711102011009910010010000100000390071011611394790100001004003840038400384003840038
102044003729900066139407025101001001000010010000500570690840018400374003738108338745101002001000020020000400374003711102011009910010010000100000620071011611394790100001004003840038400384003840038
102044003730000006139407025101001001000010010000500570690840018400374003738108338745101002001000020020000400374003711102011009910010010000100000360071011611395610100001004003840038400384003840038
10204400372990000613940702510100100100001001000050057069084001840037400373810833874510100200100002002000040037400371110201100991001001000010000050071011611394790100001004003840038400384003840038
102044003731000006139407025101001001000010010000500570690840018400374003738108838745101002001000020020000400374003711102011009910010010000100000340071011611394790100001004003840038400384003840038

1000 unrolls and 10 iterations

Result (median cycles for code): 4.0037

retire (01)cycle (02)03090b18191e1f3f4e51inst 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)9fa0a1a7a8acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10024400372990000006139407251001010100001010148505706908040018400374003738130338767100102010000202000040037400371110021109101010000100023306402162239473010000104003840038400384003840038
100244003730000003061394072510010101000010100005057069080400184003740037381303387671001020100002020000400374003711100211091010100001000049006402162239473010000104003840038400384003840038
100244003729900000061394072510010101000010100005057069080400184003740037381303387671001020100002020000400374003711100211091010100001000042006402162239473010000104003840038400384003840038
100244003730000000061394072510010101000010100005057069080400184003740037381303387671001020100002020000400374003711100211091010100001000058006402162239473010000104003840038400384003840038
1002440037299000000613940725100101010000101000050570690804001840037400373813033876710010201000020200004003740037111002110910101000010000101206402162239473010000104003840038400384003840038
100244003730000000061394072510010101000010100005057069081400184003740037381303387671001020100002020000400374003711100211091010100001000059006612162239473010000104003840038400384003840038
100244003730000000061394072510010101000010100005057069080400184003740037381303387671001020100002020000400374003711100211091010100001000012306402162239473010000104003840038400384003840038
100244003730000000061394072510010101000010100005057069080400184003740037381303387671001020100002020000400374003711100211091010100001000071206402162239473010000104003840038400384003840038
100244003730000000061394072510010101000010100005057069080400184003740037381303387671001020100002020000400374003711100211091010100001000012606402162239473010000104003840038400384003840038
100244003730000000061394072510010101000010100005057069080400184003740037381303387671001020100002020000400374003711100211091010100001000003306402162239473010000104003840038400384003840038

Test 3: Latency 1->3

Code:

  fmulx v0.8h, v1.8h, v0.8h
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 4.0037

retire (01)cycle (02)031e3a3f4e51inst 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)696b6d6edispatch 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
10204400373000061394072510100100100001001000050057069084001804003740037381083387451010020010000200200004003740037111020110099100100100001000007102161139479100001004003840038400384003840038
102044003730000613940710210100100100001001000050057069084001804003740037381083387451010020010000200200004003740037111020110099100100100001004007101161139479100001004003840038400384003840038
102044003730063061394072510100100100001001000050057069084001804003740037381083387451010020010000200200004003740037111020110099100100100001000007101161139479100001004003840038400384003840038
10204400373000061394072510100100100001001000050057069084001804003740037381083387451010020010000200200004003740037111020110099100100100001000007101161139479100001004003840038400384003840038
10204400373000061394072510100100100001001000050057069084001804003740037381083387451010020010000200200004003740037111020110099100100100001000007101161139479100001004003840038400384003840038
10204400373000061394072510100100100001001000050057069084001804003740037381083387451010020010000200200004003740037111020110099100100100001000007101161139479100001004003840038400384003840038
10204400373000061394072510100100100001001000050057069084001804003740037381083387451010020010000200200004003740037111020110099100100100001000007101161139479100001004003840038400384003840038
10204400372990061394072510100100100001001000050057069084001804003740037381083387451010020010000200200004003740037111020110099100100100001000007101160239479100001004003840038400384003840038
10204400373000061394072510100100100001001000052257069084001804003740037381083387451010020010000200200004003740037111020110099100100100001000007101161139479100001004003840038400384003840038
10204400373000061394072510100100100001001000050057069084001804003740037381083387451010020010000200200004003740037111020110099100100100001000007101161139479100001004003840038400384003840038

1000 unrolls and 10 iterations

Result (median cycles for code): 4.0037

retire (01)cycle (02)03081e3f4e51inst 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)9fa8a9accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
100244003730000613940725100101010000101000050570690814001840037400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384003840038
100244003730000613940725100101010000101000050570690814001840060400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384003840038
100244003730000613940725100101010000101000050570690814001840037400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384003840038
100244003730000613940725100101010000101000050570690814001840037400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384003840038
100244003730000613940725100101010000101000050570690814001840037400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384003840038
100244003730000613940725100101010006101000050570690814001840037400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384003840038
100244003730000613940725100101010000101000050570690814001840037400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384003840038
1002440037300007263940725100101010000101000050570690814001840037400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384008540038
100244003729900613940725100101010000101000050570690814001840037400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384003840038
100244003730000613940725100101010000101000050570690814001840037400373813033876710010201000020200004003740037111002110910101000010000640216223947310000104003840038400384003840038

Test 4: throughput

Count: 8

Code:

  fmulx v0.8h, v8.8h, v9.8h
  fmulx v1.8h, v8.8h, v9.8h
  fmulx v2.8h, v8.8h, v9.8h
  fmulx v3.8h, v8.8h, v9.8h
  fmulx v4.8h, v8.8h, v9.8h
  fmulx v5.8h, v8.8h, v9.8h
  fmulx v6.8h, v8.8h, v9.8h
  fmulx v7.8h, v8.8h, v9.8h
  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)031e1f3f51inst 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)9fa0a1a8a9accdcfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
802042004915100422580100100800001008000050064000002002120040200409973399988010020080000200160000200402004011802011009910010080000100001000511021611200370800001002004120041200412004120041
8020420040150004225801001008000010080000500640000020021200402004099733999880100200800002001600002004020040118020110099100100800001000000570511011611200370800001002004120041200412004120041
802042004015000422580100100800001008000050064000002002120040200409973399988010020080000200160000200402004011802011009910010080000100000000511011611200370800001002004120041200412004120041
80204200401500051725801001008000010080000500640000020021200402004099733999880100200800002001600002004020040118020110099100100800001000000005110116112003718800001002004120041200412004120041
802042004015000422580100100800001008000050064000002002120040200409973399988010020080000200160000200402004011802011009910010080000100001000511011611200370800001002004120041200412004120041
8020420040150004225801001008000010080000500640000020021200402004099733999880100200800002001600002004020040118020110099100100800001000000780511011611200370800001002004120041200412004120041
8020420040150120422580100100800001008000050064000002002120040200409973399988010020080000200160000200402004011802011009910010080000100002000511011611200370800001002004120041200412004120041
80204200401501203272580100100800001008000050064000002002120040200409973399988010020080000200160000200402004041802011009910010080000100002000511011611200370800001002004120041200412004120041
802042004015000422580100100800001008000050064000002010120040200409973399988010020080000200160000200402004011802011009910010080000100003000511011611200370800001002004120041200412004120041
8020420040150120422580100100800001008000050064000002002120040201459973399988010020080000200160000200402004011802011009910010080000100000030511011611200370800001002004120041200412004120041

1000 unrolls and 10 iterations

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

retire (01)cycle (02)03081e1f3f51inst 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)9fa0a8a9acc2cfd0d5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
80024200491510004125800101080000108000050640000200212004020040999631002080010208000020160000200402004011800211091010800001000000502031716171720037080000102004120041200412004120041
8002420040150000412580010108000010800005064000020021200402004099963100208001020800002016000020040200401180021109101080000100001205020081617820037080000102004120041200412004120041
80024200401500003622580010108000010800005064000020021200402004099963100208001020800002016000020040200401180021109101080000100000050200121681720037080000102004120041200412004120041
8002420040150000421258001010800001080000506400002002120040200409996310020800102080000201600002004020040118002110910108000010000005020081617620119080000102004120041200412004120041
800242004015000041258001010800001080000506400002002120040200409996310020800102080000201600002004020040118002110910108000010000005020084817820037080000102004120041200412004120041
80024200401500004125800101080000108000050640000200212004020040999631002080010208000020160000200402004011800211091010800001000000502001716171720037080000102004120041200412004120041
8002420040150000412580010108000010800005064000020021200402004099963100208001020800002016000020040200401180021109101080000100000050200171681720037080000102004120041200412004120041
800242004015000041258001010800001080000506400002002120040200409996310020800102080000201600002004020040118002110910108000010000005020081617820037080000102004120041200412004120041
8002420040150000412580010108000010800005064000020021200402004099963100208001020800002016000020040200401180021109101080000100000050200171681720037080000102004120041200412004120041
80024200401500004125800101180000108000050640000200212004020040999631002080010208000020160000200402004011800211091010800001000000502001716171720037080000102004120041200412004120041