Apple M1 Microarchitecture Research by Dougall Johnson

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

SADDW2 (vector, 8H)

Test 1: uops

Code:

  saddw2 v0.8h, v0.8h, v1.16b
  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)03090b1e3f4e51inst 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)accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)f5f6f7f8fd
100420371600061168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382038
100420371500061168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382086
100420371600061168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382038
100420371500061168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382038
100420371500061168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382038
100420371500061168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382038
100420371500061168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382038
100420371600084168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382038
100420371600061168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382038
100420371500061168725100010001000264680201820372037157231895100010002000203720371110011000073116111787100020382038203820382038

Test 2: Latency 1->2

Code:

  saddw2 v0.8h, v0.8h, v1.16b
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 2.0037

retire (01)cycle (02)031e1f3f4e51inst 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)9fa1a6a8accfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10204200371500014019687251010010010000100100005002847680120018200372003718422318745101002001000020020000200372003711102011009910010010000100000071011611197910100001002003820038200382003820038
10204200371500060819687251010010010000100100005002847680120018200372003718422318745101002001000020020000200372003711102011009910010010000100000071011611197910100001002003820038200382003820038
1020420037150006119687251010010010000100100005002847680120018200372003718422318745101002001000020020000200372003711102011009910010010000100000071011611197910100001002003820038200382003820038
1020420037150006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000071011611197910100001002003820038200382003820038
1020420037150006119687251010010010000100100005002847680120018200372003718422318745101002001000020020000200372003711102011009910010010000100000071011611197910100001002003820038200382003820038
1020420037150006119687251010010010000100100005002847680120018200372003718422318745101002001000020020000200372003711102011009910010010000100000071014211197910100001002003820038200382003820038
1020420037150006119687251010010010000100100005002847680120018200372003718422318745101002001000020020000200372003711102011009910010010000100000071011611197910100001002003820038200382003820038
1020420037150006119687251010010010000100100005002847680120018200372003718422318745101002001000020020000200372003711102011009910010010000100000071011611197910100001002003820038200382003820038
1020420037150006119687251010010010000100100005002847680120018200372003718422318745101002001000020020000200372003711102011009910010010000100000071011611197910100001002003820038200382003820038
1020420037150006119687251010010010000100100005002847680120018200372003718422318745101002001000020020000200372003711102011009910010010000100000071011611197910100001002003820038200382003820038

1000 unrolls and 10 iterations

Result (median cycles for code): 2.0037

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)9fcfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1002420037150001031968725100101010000101000050284768012001820037200371844431876710010201000020200002003720037111002110910101000010640216221978510000102003820038200382003820038
1002420037150003841968725100101010000101000050284768012001820037200371844431876710010201000020200002003720037111002110910101000010640216221978510000102003820038200382003820038
100242003715000611968725100101010000101000050284768012001820037200371844431876710010201000020200002003720037111002110910101000010668216221978510000102003820038200382003820038
100242003715000611968725100101010000101000050284768012001820037200371844431876710010201000020200002003720037111002110910101000010640216221978510000102003820038200382003820038
100242003715000611968725100101010000101000050284768012001820037200371844431876710010201000020200002003720037111002110910101000010640216221978510000102003820038200382003820038
100242003715000611968725100101010000101000050284768012001820037200371844431876710010201000020200002003720037111002110910101000010640216221978510000102003820038200382003820038
100242003715000611968725100101010000101000050284768012001820037200371844431876710010201000020200002003720037111002110910101000010640216221978510000102003820038200382003820038
10024200371500186611968725100101010000101000050284768012001820037200371844431876710010201000020200002003720037111002110910101000010640216221978510000102003820038200382003820038
100242003715000611968725100101010000101000050284768002001820037200371844431876710010201000020200002003720037111002110910101000010640216221978510000102003820038200382003820038
1002420037150003021968725100101010000101000050284768012001820037200371844431876710010201000020200002003720037111002110910101000010640216221978510000102003820038200382003820038

Test 3: Latency 1->3

Code:

  saddw2 v0.8h, v1.8h, v0.16b
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 2.0037

retire (01)cycle (02)03181e3a3f4e51inst 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)9fc2cfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
102042003715000061196872510100100100001001000050028476802001820037200371842231874510100200100002002000020037200371110201100991001001000010007101161119791100001002003820038200382003820038
102042003715000061196872510100100100001001000050028476802001820037200371842231874510100200100002002000020037200371110201100991001001000010007101161119791100001002003820038200382003820038
1020420037150000346196872510100100100001001000050028476802001820037200371842231874510100200100002002062220084200371110201100991001001000010007101161119791100001002003820038200382003820038
1020420037150000103196872510100100100001001000050028476802001820037200371842231874510100200100002002000020037200371110201100991001001000010007101161119791100001002003820038200382003820038
102042003715000061196872510100100100001001000050028476802001820037200371842231874510100200100002002000020037200371110201100991001001000010007101161119791100001002003820038200382003820038
102042003715000061196872510100100100001001000050028476802001820037200371842231874510100200100002002000020037200371110201100991001001000010007101161119791100001002003820038200382003820038
1020420037150000631196872510100100100001001000050028476802001820037200371842231874510100200100002002000020037200371110201100991001001000010007101161119791100001002003820038200382003820038
1020420037150000124196872510100100100001001000050028476802001820037200371842231874510100200100002002000020037200371110201100991001001000010007101161119791100001002003820038200382003820038
102042003714900061196872510100100100001001000050028476802001820037200371842231874510100200100002002000020037200371110201100991001001000010007101161119791100001002003820038200382003820038
102042003715010061196872510100100100001001000050028476802001820037200371842231874510100200100002002000020037200371110201100991001001000010007101161119791100001002003820038200382003820038

1000 unrolls and 10 iterations

Result (median cycles for code): 2.0037

retire (01)cycle (02)0318191e1f3f4e51inst 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
1002420037150001061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000061196872510010101000010100005028476800200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000061196872510010101000010100005028476800200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000061196872510010101000010100005028476800200182003720037184443187671001020100002020000200372003731100211091010100001000000006402162219785010000102008520086200382003820038
10024200371500030251196872510010101000010100005028476800200182003720037184443187671001020100002020000200372003711100211091010100001000000006402161219785010000102003820038200382003820038
1002420037150000061196872510010101000010100005028476800200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000061196872510010101000010100005028476800200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000061196872510010101000010100005028476800200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000061196872510010101001210100005028476800200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000061196872510010101000010100005028476800200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038

Test 4: throughput

Count: 8

Code:

  saddw2 v0.8h, v8.8h, v9.16b
  saddw2 v1.8h, v8.8h, v9.16b
  saddw2 v2.8h, v8.8h, v9.16b
  saddw2 v3.8h, v8.8h, v9.16b
  saddw2 v4.8h, v8.8h, v9.16b
  saddw2 v5.8h, v8.8h, v9.16b
  saddw2 v6.8h, v8.8h, v9.16b
  saddw2 v7.8h, v8.8h, v9.16b
  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)0308181e1f3f51inst 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)9fa0a8accfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
80204200381500000402580100100800001008000050064000002001920038200389973399968010020080000200160000200382003811802011009910010080000100000511021611200350800001002003920039200392003920039
80204200381500000402580100100800001008000050064000002001920038200389973399968010020080000200160000200382003811802011009910010080000100000511011611200350800001002003920039200392003920039
80204200381500000402580100100800001008000050064000002001920038200389973399968010020080000200160000200382003811802011009910010080000100000511011611200350800001002003920039200392003920039
80204200381500000402580100100800001008000050064000002001920038200389973399968010020080000200160000200382003811802011009910010080000100000511011611200350800001002003920039200392003920039
802042003815000007052580100100800001008000050064000002001920038200389973399968010020080000200160000200382003811802011009910010080000100000511011611200350800001002003920039200392003920039
802042003815000180402580100100800001008000050064000002001920038200389973399968010020080000200160000200382003811802011009910010080000100000511011611200350800001002003920039200392003920039
80204200381500000402580100100800001008000050064000002001920038200389973399968010020080000200160000200382003811802011009910010080000100000511011611200350800001002003920039200392003920039
802042003815000004025801001008000010080000500640000020019200382003899733999680100200800002001600002003820038118020110099100100800001000005110116112003515800001002003920039200392003920039
80204200381500030402580100100800001008000050064000002001920038200389973399968010020080000200160000200382003811802011009910010080000100000511011611200350800001002003920039200392003920039
80204200381500000402580100100800001008000050064000002001920038200389973399968010020080000200160000200382003811802011009910010080000100000511011611200350800001002003920039200392003920039

1000 unrolls and 10 iterations

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

retire (01)cycle (02)0307080e191e3a3f51inst 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)9fa7a8a9accdcfd5d6ddinst fetch restart (de)e0? int output thing (e9)ea? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
8002420048150002000392580010108000010800005064000012001920038200389996310018800102080000201600002003820038118002110910108000010000005020716107200350080000102003920039200392003920039
800242003815000000039258001010800001080000506400001200192003820038999631001880010208000020160000200382003811800211091010800001000000502091695200350080000102003920039200392003920039
800242003815000000039258001010800001080000506400000200192003820038999631001880010208000020160000200382003811800211091010800001000000502071699200350080000102003920039200392003920039
800242003815000000039258001010800001080000506400000200192003820038999631001880010208000020160000200382003811800211091010800001000000502071667200350080000102003920039200392003920039
800242003815000000060258001010800001080000506400000200192003820038999631001880010208000020160000200382003811800211091010800001000000502061678200350080000102003920039200392003920039
800242003815000000039258001010800001080000506400001200192003820038999631001880010208000020160000200382003811800211091010800001000000502091667200350080000102003920039200392003920039
800242003815000000039258001010800001080000506400001200192003820038999631004680010208000020160000200382003811800211091010800001000000502091689200350080000102003920039200392003920039
800242003815000000039258001010800001080000506400001200192003820038999631001880010208000020160000200382003811800211091010800001001030502071676200350080000102003920039200392003920039
800242003815000000039258001010800001080000506400001200192003820038999631001880010208000020160000200382003811800211091010800001000000502091697200350080000102003920039200392003920039
8002420038150000090392580010108000010800005064000002001920038200389996310018800102080000201600002003820038118002110910108000010000005020916972003501680000102003920039200392003920039