Apple M1 Microarchitecture Research by Dougall Johnson

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

SUB (vector, 4H)

Test 1: uops

Code:

  sub v0.4h, v0.4h, v1.4h
  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)0318191e3f4e51inst 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
10042037150006116872510001000100026468012018203720371572318951000100020002037203711100110000073116111787100020382038203820382038
10042037150006116872510001000100026468002018203720371572318951000100020002037203711100110000073116111787100020382038203820382038
10042037150006116872510001000100026468002018203720371572318951000100020002037203711100110000073116111787100020382038203820382038
10042037150008216872510001000100026468002018203720371572318951000100020002037203711100110000073116111787100020382038203820382038
10042037150006116872510001000100026468002018203720371572318951000100020002037203711100110000073116111787100020382038203820382038
100420371500361168725100010001000264680020182037203715723189510001000200020372037111001100001873116111787100020382038203820382038
10042037150006116872510001000100026468002018203720371572318951000100020002037203711100110000073116111787100020382038203820382038
10042037150006116872510001000100026468012018203720371572318951000100020002037203711100110000073116111787100020382038203820382038
10042037150006116872510001000100026468012018203720371572318951000100020002037203711100110000073116111787100020382038203820382038
10042037160006116872510001000100026468012018203720371572318951000100020002037203711100110000073116111787100020382038203820382038

Test 2: Latency 1->2

Code:

  sub v0.4h, v0.4h, v1.4h
  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)03080b18191e1f3a3f4e51inst 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)9fa0a1a6a7a8a9acc2c5branch mispredict (cb)cdcfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10204200371500000225006119687251010010010000100100005002847680200182003720037184296187401010020010008200200162003720037111020110099100100100001000000000011171801600198010100001002003820038200382003820038
1020420037150000018006119687251010010010000100100005002847680200182003720037184296187411010020010008200200162003720037111020110099100100100001000000000000071011611197910100001002003820038200382003820038
102042003715000000006119687251010010010000100100005002847680200182003720037184223187451010020010000200200002003720037111020110099100100100001000000000000071011611197910100001002003820038200382003820038
102042003715000000006119687251010010010000100100005002847680200182003720037184223187611010020010000200200002003720037111020110099100100100001000000000000071011621197910100001002003820038200382003820038
102042003715000000006119687251010010010000100100005002847680200182003720037184223187451010020010000200200002003720037111020110099100100100001000000100000071011611197910100001002003820038200382003820038
1020420037150000054006119687251010010010000100100005002847680200182003720037184223187451010020010000200200002003720037111020110099100100100001000000000000071011611197910100001002003820038200382003820038
102042003715000000006119687251010010010000100100005002847680200182003720037184223187451010020010000200200002003720037111020110099100100100001000000000000071011611197910100001002003820038200382003820038
102042003715000000006119687251010010010000100100005002847680200182003720037184223187451010020010000200200002003720037111020210099100100100001000000000000071011611197910100001002003820038200382003820038
1020420037150000027006119687251010010010000100100005002847680200182003720037184223187451010020010000200200002003720037111020110099100100100001000000000000071011611197910100001002003820038200382003820038
102042003715000000006119687251010010010000100100005002847680200182003720037184223187451010020010000200200002003720037111020110099100100100001000000000000071011611197910100001002003820038200382003820038

1000 unrolls and 10 iterations

Result (median cycles for code): 2.0037

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)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)9fa0a1a7a8a9acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
100242003715000000210061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000006404162219785010000102003820038200382003820038
1002420037150000000061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
100242003715000000249061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
10024200371500000027061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000000061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
100242003715000000363061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000000061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420037150000000061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000006402162219785010000102003820038200382003820038
1002420084150000000061196872510010101000010100005028476801200182003720037184443187671001020100002020000200372003711100211091010100001000000007273572219931310000102027320227202782022820181
100242022715110154540352205119632118100771110060131060860285409512009020274202261846024188591077322106712021670202262027261100211091010100001004120792827624492319967110000102027420274202752022920227

Test 3: Latency 1->3

Code:

  sub v0.4h, v1.4h, v0.4h
  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)03080b18191e1f3a3f4e51inst 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)9fa1a6a7a8a9acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1020420037150000057006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000000071011611197910100001002003820038200382003820038
102042003715000000006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000000071011611197910100001002003820038200382003820038
1020420037150000012006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000000071011611197910100001002003820038200382003820038
102042003715000006006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003721102011009910010010000100000000071011611197910100001002003820038200382003820038
102042003715000000006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000000071011611197910100001002003820038200382003820038
1020420037150000027006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000000071011611197910100001002003820038200382003820038
1020420037150000012006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000000071011611197910100001002003820038200382003820038
102042003715000000006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000000071011611197910100001002003820038200382003820038
1020420037150000036006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000000071011611197910100001002003820038200382003820038
10204200371500000291006119687251010010010000100100005002847680020018200372003718422318745101002001000020020000200372003711102011009910010010000100000000071011611197910100001002003820038200382003820038

1000 unrolls and 10 iterations

Result (median cycles for code): 2.0037

retire (01)cycle (02)030818191e3f4e51inst 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)9fa8accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
100242003715000006119687251001010100001010000502847680200182003720037184443187671001020100002020000200372003711100211091010100001000640216221978510000102003820038200382003820038
100242003715000006119687251001010100001010000502847680200182003720037184443187671001020100002020000200372003711100211091010100001000640216221978510000102003820038200382003820038
1002420037150000216119687251001010100001010000502847680200182003720037184443187671001020100002020000200372003711100211091010100001000640216221978510000102003820038200382003820038
1002420037150000186119687251001010100001010000502847680200182003720037184443187671001020100002020000200372003711100211091010100001000640216221978510000102003820038200382003820038
100242003715000066119687251001010100001010000502847680200182003720037184443187671001020100002020000200372003711100211091010100001000640216221978510000102003820038200382003820038
1002420037150000276119687251001010100001010000502847680200182003720037184443187671001020100002020000200372003711100211091010100001000640216221978510000102003820038200382003820038
100242003715000006119687251001010100001010000502847680200182003720037184443187671001020100002020000200372003711100211091010100001000640216221978510000102003820038200382003820038
100242003715000006119687251001010100001010000502847680200182003720037184443187671001020100002020000200372003711100211091010100001000640216221978510000102003820038200382003820038
100242003715000006119687251001010100001010000502847680200182003720037184443187671001020100002020000200372003711100211091010100001000640216221978510000102003820038200382003820038
100242003715000006119687251001010100001010000502847680200182003720037184443187671001020100002020998200372008411100211091010100001000640216221978510000102003820038200382003820038

Test 4: throughput

Count: 8

Code:

  sub v0.4h, v8.4h, v9.4h
  sub v1.4h, v8.4h, v9.4h
  sub v2.4h, v8.4h, v9.4h
  sub v3.4h, v8.4h, v9.4h
  sub v4.4h, v8.4h, v9.4h
  sub v5.4h, v8.4h, v9.4h
  sub v6.4h, v8.4h, v9.4h
  sub v7.4h, v8.4h, v9.4h
  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)03080b18191e1f3f51inst 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)9fa0a1a6a7a8a9acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
80204200481500000904025801001008000010080000500640000200192003820038997339996801002008000020016000020038200381180201100991001008000010000000000511031611200350800001002003920039200392003920039
80204200381500000004025801001008000010080000500640000200192003820038997339996801002008000020016000020038200381180201100991001008000010020000000511011611200350800001002003920039200392003920039
802042003815000009012225801001008000010080000500640000200192003820038997339996801002008000020016000020038200381180201100991001008000010000000000511011611200350800001002003920039200392003920039
802042003815000001804025801001008000010080000500640000200192003820038997339996801002008000020016000020038200381180201100991001008000010000000000511011612200350800001002003920039200392003920039
802042003815000002704025801001008000010080000626640000200192003820038997339996801002008000020016000020038200381180201100991001008000010000000000511011611200350800001002003920039200392003920039
802042003815000002404025801001008000010080000500640000200192003820038997339996801002008009620016000020038200381180201100991001008000010000000000511011611200350800001002003920039200392003920039
8020420038150000051081925801001008000010080000500640000200192003820038997339996801002008000020016000020038200381180201100991001008000010000000000511011611200780800001002003920039200392003920039
802042003815000001504025801001008000010080000500640000200192003820038997339996801002008000020016000020038200381180201100991001008000010000000000511021611200350800001002003920039200392003920039
802042003815000001504025801001008000010080000500640000200192003820038997339996801002008000020016000020038200381180201100991001008000010000000030511011611200350800001002003920039200392003920039
8020420038150000040206125801001008000010080000500640000200192003820038997339996801002008000020016000020038200381180201100991001008000010000000000511011611200350800001002003920039200392003920039

1000 unrolls and 10 iterations

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

retire (01)cycle (02)030418191e1f3f51inst 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)9fa8accficache miss (d3)itlb miss (d4)d5d6d9daddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
800242004815000033039258001010800001080000506400000200192003820038999631001880010208000020160000200382003811800211091010800001000502101171600171720035080000102003920039200392003920039
80024200381501000039258001010800001080000506400001200192003820038999631001880010208000020160000200382003811800211091010800001000502101155000141620035080000102003920039200392003920039
80024200381501002703925800101080000108000050640000020019200382003899963100188001020800002016000020038200381180021109101080000100350210114160081720035080000102003920039200392003920039
800242003815010015039258001010800001080000506400001200192003820038999631001880010208000020160000200382003811800211091010800001000502101171600171320035080000102003920039200392003920039
800242003815010027039258001010800001080000506400000200192003820038999631001880010208000020160000200382003811800211091010800001000502101101600181620035080000102003920039200392003920039
800242003815010015039258001010800001080000506400000200192003820038999631001880010208000020160000200382003811800211091010800001000502101171600141720035080000102003920039200392003920039
800242003815010015039258001010800001080000506400001200192003820038999631001880010208000020160000200382003811800211091010800001000504601161600141720035080000102003920039200392003920039
800242003815010066039258001010800001080000506400001200192003820038999631001880010208000020160000200382003811800211091010800001000502101141600171420035080000102003920039200392003920039
800242003815010015039258001010800001080000506400001200192003820038999631001880010208000020160000200382003811800211091010800001000502101171600141720035080000102003920039200392003920039
80024200381501000039258001010800001080000506400000200192003820038999631001880010208000020160000200382003811800211091010800001000502141151600131820035080000102003920039200392003920039