Apple M1 Microarchitecture Research by Dougall Johnson

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

FNMSUB (scalar, S)

Test 1: uops

Code:

  fnmsub s0, s0, s1, s2
  movi v0.16b, 1
  movi v1.16b, 2
  movi v2.16b, 3

(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)0308090a0b18191e1f3f4e51inst 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)a1a8a9acc2cfd5d6ddinst fetch restart (de)e0? fp/simd (ee)f5f6f7f8fd
1004403730000000008234072510001000100053190814018403740373258338951000100030004037403711100110000000073216113473100040384038403840384038
1004403731000000006134072510001000100053190804018403740373258338951000100030004037403711100110000000073116113473100040384038403840384038
1004403730000000006134072510001000100053190814018403740373258338951000100030004037403711100110000003073116113473100040384038403840384038
1004403730000000006134072510001000100053190804018403740373258338951000100030004037403711100110000000073116113473100040384038403840384038
1004403730000000006134072510001000100053190814018403740373258338991000100030004037403711100110000000073116113473100040384038403840384038
1004403731000000006134072510001000114853190804018403740373258338951000100030004037403711100110000000073116113473100040384038403840384038
1004403730000000006134072510001000100053190814018403740373258338951000100030004037403711100110000000073132113473100040384038403840384038
10044037300000005406134072510001000100053190804018403740373258338951000100030004037403711100110000100073116113473100040384038403840384038
10044037300000000012434072510001000100053190804018403740373258338951000100030004037403711100110000000073116113473100040384038403840384038
1004403730000000006134072510001000100053190804018403740373258338951000100030004037403711100110000000073116113473100040384038403840384038

Test 2: Latency 1->2

Code:

  fnmsub s0, s0, s1, s2
  movi v0.16b, 1
  movi v1.16b, 2
  movi v2.16b, 3

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 4.0037

retire (01)cycle (02)03081e3a3f4e51inst 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)9fa8accfd0d5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
102044003730000061394072510100100100001001000050057069080400184003740037381083387451010020010000200300004003740037111020110099100100100001000671002162239479100001004008440038400384003840038
102044003730000061394072510100100100001001000050057069081400184003740037381083387451010020010000200300004003740037111020110099100100100001000071012162239479100001004003840038400384003840038
102044003730000082394072510100100100001001000050057069080400184003740037381083387451010020010000200300004003740037111020110099100100100001000071012162239479100001004003840038400384003840038
1020440037300000115394072510100100100001001000050057069080400184003740037381083387451010020010000200300004003740037111020110099100100100001000071012162239479100001004003840038400384003840038
102044003730000061394072510100100100001001000050057069080400184003740037381083387451010020010000200300004003740037111020110099100100100001000071012162239553100001004003840038400384003840038
102054003730000061394072510100100100001001000050057069081400184003740037381083387451010020010000200300004003740037111020110099100100100001000071012163239479100001004003840038400384003840038
102044003730000061394072510100100100001001000050057069081400184003740037381083387451010020010000200300004003740037111020110099100100100001000071012162239479100001004003840038400384003840038
102044003729900061394072510100100100001001000050057069080400184003740037381083387451010020010000200300004003740037111020110099100100100001000071012162239479100001004003840038400384003840038
10204400372990005063940725101001001000010010000500570690814001840037400373810833874510100200100002003000040037400371110201100991001001000010025971012162239479100001004003840038400384003840038
1020440037300000458394072510100100100001001000050057069080400184003740037381083387451010020010000200300004003740037111020110099100100100001000071012162239479100001004003840038400384003840038

1000 unrolls and 10 iterations

Result (median cycles for code): 4.0037

retire (01)cycle (02)03191e3f4e51inst 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)9fa8accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10024400373000061394072510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001000640216223947310000104003840038400384003840038
100244003729900232394072510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001000640216223947310000104003840038400384003840038
10024400372990961394072510010101000010100005057069080400184003740037381303387671001020100002030000400374003711100211091010100001000640216223947310000104003840038400384003840038
1002440037299013261394072510010101000010100005057069080400534003740037381303387671001020100002030000400374003711100211091010100001003640216223947310000104008540131400384003840038
10024400372990061394074510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001000640216223947310000104003840038400384003840038
100244003730001261394072510010101000010100005057069080400184003740037381303387671001020100002030000400374003711100211091010100001000640216223947310000104003840038400384003840038
100244003730001261394072510010101000010100005057069080400184022740037381303387671001020100002030000400374003711100211091010100001000640216223947310000104003840038400384003840038
10024400373000061394072510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001000640216223947310000104003840038400384003840038
10024400373000061394072510010101000010100005057069080400184003740037381303387671001020100002030000400374003711100211091010100001000640216223947310000104003840038400384003840038
10024400373000061394072510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001000640216223947310000104003840038400384003840038

Test 3: Latency 1->3

Code:

  fnmsub s0, s1, s0, s2
  movi v0.16b, 1
  movi v1.16b, 2
  movi v2.16b, 3

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 4.0037

retire (01)cycle (02)0308090b18191e1f3a3f4e51inst 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)9fa1a8acc2c5cfd0d5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
102044003730000000000613940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000000710031622394790100001004003840038400384003840038
102044003730000000000613940725101001001000010010000500570690814001840037400373810833874510100200100002003000040037400371110201100991001001000010000000710021622394790100001004003840038400384003840038
102044003730000000000613940725101001001000010010000500570690814001840037400373810833874510100200100002003000040037400371110201100991001001000010000000710021622394790100001004003840038400384003840038
102044003729900000000613940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000000710121622394790100001004003840038400384003840038
102044003729900000000613940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000000710121622394790100001004003840038400384003840038
1020440037300000000001033940725101001001000010010000500570690814001840037400373810833874510100200100002003000040037400371110201100991001001000010000000710121622394790100001004003840038400384003840038
102044003730000000000613940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000000710121622394790100001004003840038400384003840038
102044003729900000000613940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400841110201100991001001000010000000710121622394790100001004003840038400384003840038
102044003730000000000843940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000300710121622394790100001004003840038400384003840038
102044003729910000000613940725101001001000010010148500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000000710121622394790100001004003840038400384003840038

1000 unrolls and 10 iterations

Result (median cycles for code): 4.0037

retire (01)cycle (02)03041e1f3f4e51inst 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)9fa1a8acc2cfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10024400373001006139407251001010100001010000505706908140018400374003738130338767100102010000203000040037400371110021109101010000100930640316223947310000104003840038400384003840038
100244003730000061394072510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001002700640216223947310000104003840038400384003840038
100244003730000061394072510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001001900640216223947310000104003840038400384003840038
100244003730000061394072510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001002700640216223947310000104003840038400384003840038
10024400372990006139407251001010100001010000505706908140018400374003738130338767100102010000203000040037400371110021109101010000100130640216223947310000104003840038400384003840038
100244022430000061394072510016101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001004700640216223947310000104003840038400384003840038
100244003729900061394072510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001006100640216233947310000104003840038400384003840038
100244003730000061394072510010101000010100005057069081400184003740037381303387671001020100002030000400374003711100211091010100001003600640216223947310000104003840038400384003840038
10024400373000006139407251001010100001010000505706908140018400374003738130338767100102010000203000040037400371110021109101010000100160640216223947310000104003840038400384003840038
10024400372990006139407251001010100001010000505706908140018400374003738130338767100102010000203000040037400371110021109101010000100760640216223947310000104003840038400384003840038

Test 4: Latency 1->4

Code:

  fnmsub s0, s1, s2, s0
  movi v0.16b, 1
  movi v1.16b, 2
  movi v2.16b, 3

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 4.0037

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)9fa8a9accfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
102044003729900019206139407251010010010000100100005005706908040018400374003738112338763101002001000020030000400374003711102011009910010010000100512671041623394790100001004003840038400864003840038
102044003729900003521873940725101001001000010010000500570690804001840037400373810833874510100200100002003099940037400371110201100991001001000010030671021622394790100001004003840038400384003840038
10204400373000001202293940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000671021622394790100001004003840038400384003840038
1020440037300000002083940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000071021622394790100001004003840038400384003840038
1020440037300000001683940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000071021622394790100001004003840038400384003840038
1020440037299000001703940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000071021622394790100001004003840038400384003840038
102044003730003000613940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000071021622394790100001004003840038400384003840038
1020440037300000005863940725101001001000010010000500570690804001840037400373810833874510100200100002003000040037400371110201100991001001000010000071021622394790100001004003840038400384003840038
1020440037300000001663940725101001001000010010000500570690804001840037400373810833874510100200104892003000040037400371110201100991001001000010000071021622394790100001004003840038400384003840038
1020440037299000005583940725101001001000010010000500570690804001840230400373810833874510100200100002003000040037400371110201100991001001000010000071021623394790100001004003840038400384003840038

1000 unrolls and 10 iterations

Result (median cycles for code): 4.0037

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)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)9fa8a9accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1002440037300000613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010000640316333947310000104003840038400384003840038
1002440037300006613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010000640316333947310000104003840038400384003840038
1002440037300000613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010003640316333947310000104003840038400384003840038
1002440037300000613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010000640316433947310000104003840038400384003840038
1002440037300000613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010000640316333947310000104003840038400384003840038
1002440037300000613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010000640316333947310000104003840038400384003840038
1002440037300000613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010000640316333947310000104003840038400384003840038
1002440037299000613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010000640316333947310000104003840038400384003840038
1002440037300000613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010000640316333947310000104003840038400384003840038
1002440037300000613940725100101010000101000050570690840018400374003738130033876710010201000020300004003740037111002110910101000010000640316433947310000104003840038400384003840038

Test 5: throughput

Count: 8

Code:

  fnmsub s0, s8, s9, s10
  fnmsub s1, s8, s9, s10
  fnmsub s2, s8, s9, s10
  fnmsub s3, s8, s9, s10
  fnmsub s4, s8, s9, s10
  fnmsub s5, s8, s9, s10
  fnmsub s6, s8, s9, s10
  fnmsub s7, s8, s9, s10
  movi v8.16b, 9
  movi v9.16b, 10
  movi v10.16b, 11

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

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

retire (01)cycle (02)03080b1e3f51inst 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)9fa8accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
80204200601500004225801001008000010080000500640000120021200402004099733999880100200800002002400002004020040118020110099100100800001000051101161120037800001002004120041200412004120041
80204200401500004225801001008000010080000500640000020021200402004099733999880100200800002002400002004020040118020110099100100800001000051101161120037800001002004120041200412004120041
80204200931501104225801001008000010080000500640000020021200402004099733999880100200800002002400002004020040118020110099100100800001000051101161120037800001002004120041200412004120041
80204200401500004225801001008000010080000500640000020021200402004099733999880100200800002002400002004020040118020110099100100800001000051101161120037800001002004120041200412004120041
80204200401500004225801001008000010080000500640000020021200402004099733999880100200800002002400002004020040118020110099100100800001000051101161120037800001002004120041200412004120041
80204200401500004225801001008000010080000500640000020021200402004099733999880100200800002002400002004020040118020110099100100800001002051101161120037800001002004120041200412004120041
80204200401500004225801001008000010080000500640000120021200402004099733999880100200800002002400002004020040118020110099100100800001000051101161120037800001002004120041200412004120041
802042004015000042258010010080000100800005006400000200212004020040997325999880100200800002002400002004020040118020110099100100800001000651101161120037800001002004120041200412004120041
80204200401500004225801001008000010080000500640000020021200402004099733999880100200800002002400002004020040118020110099100100800001000051101161120037800001002004120041200412004120041
802042004015000061225801001008000010080000500640000020021200402004099733999880100200800002002400002004020040118020110099100100800001000051101161120037800001002004120041200412004120041

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)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)9fa1a8a9acc2cfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
80024200501500004125800101080000108000050640000012002120040200409996310020800102080000202400002004020040118002110910108000010000005020316342003780000102004120041200412004120041
80024200401500004125800101080000108000050640000012002120040200409996310020800102080000202400002004020040118002110910108000010000005020316322003780000102004120041200412004120041
80024200401500004125800101080000108000050640000012002120040200409996310020800102080000202400002004020040118002110910108000010000005020316352003780000102004120041200412004120041
80024200401500004125800101080000108000050640000002002120040200409996310020800102080000202400002004020040118002110910108000010000005020316332003780000102004120041200412004120041
800242004015000104125258011010801031080000506400001120021200402004010005710048801142080000202400002004020040118002110910108000010200025020316332033380000102010320096200412004120101
8002420040150012012525800101080000108000050640000012002120040200409996310020800102080000202400002004020040118002110910108000010000005020316332003780000102004120041200412004120041
80024200401500004125800101080000108000050640000002002120040200409996310020800102080000202400002004020040118002110910108000010000005020316332003780000102004120041200412004120041
80024200401500004125800101080000108000050640000002002120040200409996310020800102080000202400002004020040118002110910108000010000005020316542003780000102004120041200412004120041
80024200401500004125800101080000108000050640000002002120040200409996310020800102080000202400002004020040118002110910108000010000005020316232003780000102004120041200412004120041
800242004015000032625800101080000108000050640000012002120040200409996310020800102080000202400002004020040118002110910108000010000005020316352003780000102004120041200412004120041