Apple M1 Microarchitecture Research by Dougall Johnson

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

FCCMPE (scalar, S)

Test 1: uops

Code:

  fccmpe s0, s1, #0, lt
  mov x0, 1
  mov x1, 2
  mov x2, 3
  mov x3, 4

(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)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)9fa6a8a9accfd5d6ddinst fetch restart (de)e0f5f6f7f8fd
1004203716006193725100010001000689841201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038
1004203716006193725100010001000689841201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038
1004203715006193725100010001000689841201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038
1004203715006193725100010001000689841201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038
1004203715006193725100010001000689841201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038
1004203715006193725100010001000689840201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038
10042037150010393725100010001000689841201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038
1004203715006193725100010001000689840201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038
1004203715006193725100010001000689840201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038
1004203715006193725100010001000689841201820372037178831895100010003000203720371110011000100000007311611196920382038203820382038

Test 2: Latency 3->1

Chain cycles: 2

Code:

  fccmpe s0, s1, #0, lt
  fcsel d0, d2, d3, eq
  mov x0, 1
  mov x1, 2
  mov x2, 3
  mov x3, 4

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code, minus 2 chain cycles): 2.0037

retire (01)cycle (02)030b181e1f3f4e51inst 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)9fa0a7a8acc5branch mispredict (cb)cdcfd0d2d5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
202044003730000006139822252010010020000100200005002853880154001840037400373732363749120100200200082006002440037400371120201100991001002000020100000011113175101600399850100001004003840038400384003840038
202044003730000006139822252010010020000100200005002853880004001840037400373732363749120100200200082006002440037400371120201100991001002000020100000011113170001600399190100001004003840038400384003840038
2020440037300000061398222520100100200001002000050028538800540018400374003737316337495201002002000020060000400374003711202011009910010020000201000000000131051116113990926100001004003840038400384003840038
202044003729900006139822252010010020000100200005002853880154001840037400373731633749520100200200002006000040037400371120201100991001002000020100000000013105111611399090100001004003840038400384003840038
202044003729900006139822252010010020000100200005002853880154001840037400373731633749520100200200002006000040037400371120201100991001002000020100000000013105111611399090100001004003840038400384003840038
202044003729900006139822252010010020000100200005002853880154001840037400373731633749520100200200002006000040037400371120201100991001002000020100000000013105111611399090100001004003840038400384003840038
202044003730000006139822252010010020000100200005002853880154001840037400373731633749520100200200002006000040037400371120201100991001002000020100000000013105111611399090100001004003840038400384003840038
20204400373000000115139822252010010020000100200005002853880104001840037400373731633749520100200200002006000040037400831120201100991001002000020100210000013350011611399090100001004003840038400384003840038
202044003730000006139822252010010020000100200005002853880004001840037400373731633749520100200200002006000040037400371120201100991001002000020100000000013100011611399090100001004003840038400384003840038
2020440037300001206139822252010010020000100200005002853880004001840037400373731633749520100200200002006000040037400371120201100991001002000020100000300013100011611399090100001004003840038400384003840038

1000 unrolls and 10 iterations

Result (median cycles for code, minus 2 chain cycles): 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)9fa8accfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
200244003730000006139822252001010200001020000502853880140018400374003737338337517200102020000206000040084400841120021109101020000200100012704163339910010000104003840038400384003840038
2002440037300000012639822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100012703164339910010000104003840038400384003840038
200244003730000006139822252001010200001020000502853880040018400374003737338337517200102020000206000040037400371120021109101020000200100012703163439910010000104003840038400384003840038
200244003730000006139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100012703163339910010000104003840038400384003840038
2002440037299000015639822252001010200001020000502853880040018400374003737338337517200102020000206000040037400371120021109101020000200100012703163339910010000104003840038400384003840038
200244003729900006139822252001010200001020000502853880040018400374003737338337517200102020000206000040037400371120021109101020000200100012703163339910010000104003840038400384003840038
200244003730000006139822252001010200001020000502853880040018400374003737338337517200102020000206000040037400371120021109101020000200100012703163339910010000104003840038400384003840038
200244003730000006139822252001010200001020000502853880040018400374003737338337517200102020000206000040037400371120021109101020000200100012703163339910010000104003840038400384003840038
20024400373000000613982225200101020000102000050285535314001840037400373733833751720010202000020600004003740037112002110910102000020010021312703163339910010000104003840038400384003840038
2002440037300003025139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100012703163339910010000104003840038400384003840038

Test 3: Latency 3->2

Chain cycles: 2

Code:

  fccmpe s0, s1, #0, lt
  fcsel d1, d2, d3, eq
  mov x0, 1
  mov x1, 2
  mov x2, 3
  mov x3, 4

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code, minus 2 chain cycles): 2.0037

retire (01)cycle (02)030b1e1f3a3f4e51inst 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)9faccfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
202044003730004500613982225201001002000010020000500285388014001840037400373731633749520100200200002006000040037400371120201100991001002000020100013105161139909100001004003840038400384003840038
202044003730002400613982225201001002002410020000500285388014001840178400373731633749520100200200002006000040037400371120201100991001002000020100013103161139909100001004003840038400864008640038
2020440037299072880613982225201001002004810020000500285388014001840037400373731633749520100200200002006000040037400371120201100991001002000020100013101161139909100001004003840038400384003840038
2020440037300052800613982225201001002000010020000500285388014001840037400373731633749520100202200002006000040037400371120201100991001002000020100013101161139909100001004003840038400384003840038
2020440037300024007263982225201001002000010020000500285388014001840037400373731633749520100200200002006000040037400371120201100991001002000020100013101161139909100001004003840038400384003840038
2020440037300036006139822252010010020000100200005002853880140018400374003737316337495201002002000020060000400374003711202011009910010020000201001213781161139909100001004003840038400384003840038
202044003730002401613982225201001002000010020000500285388014001840037400373731633749520100200200002006000040037400371120201100991001002000020100013101161139909100001004003840229400384003840038
202044003730003900613982225201001002000010020000500285388004001840037400373731633749520100200200002006000040037400371120201100991001002000020100013101161139909100001004003840038400384003840038
2020440037300014400613982225201001002000010020000500285388014001840037400373731633749520100200200002006000040037400371120201100991001002000020100013101161139909100001004003840038400384003840038
20204400373000000613982225201001002000010020000500285388014001840037400373731633749520100200200002006000040037400371120201100991001002000020100013101161139909100001004003840038400384003840038

1000 unrolls and 10 iterations

Result (median cycles for code, minus 2 chain cycles): 2.0037

retire (01)cycle (02)03080b18191e3f4e51inst 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)9fa1a6a8a9accfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
2002440037300000006139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100000012701161139910010000104003840038400384003840038
2002440037300000006139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100000012701161439910010000104003840038400384003840038
2002440037300000006139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100000012701161139910010000104003840038400384003840038
20024400373000000015639822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100000312701161139910010000104003840038400384003840038
20024400373000001276139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100010012704161139910010000104003840038400384003840038
2002440037300000006139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100000012701161139910010000104003840038400384003840038
2002440037300000006139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100000012701161139910010000104003840038400384003840038
2002440037299000006139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100000012701164139910010000104003840038400384003840038
20024400373000000246139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100100012701161139910010000104003840038400384003840038
20024400373000010156139822252001010200001020000502853880140018400374003737338337517200102020000206000040037400371120021109101020000200100000012701161139910010000104003840038400384003840038

Test 4: Latency 3->3

Code:

  fccmpe s0, s1, #0, lt
  mov x0, 1
  mov x1, 2
  mov x2, 3
  mov x3, 4

(non-fused SUB/CBNZ loop)

100 unrolls and 100 iterations

Result (median cycles for code): 2.0037

retire (01)cycle (02)03081e1f3f4e51inst 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 integer (97)inst neon or fp (9a)9fa6a8a9accfd0d5d6ddinst fetch restart (de)e0? int output thing (e9)gpr retires (ef)f5f6f7f8fd
10204200371500003599937251020020010000200100001100707984120018200372003718638318745102002001000020030000200372003711102011009910010000100000000710031633199691001002003820038200382003820038
1020420037150000619937251020020010000200100001100707984120018200372003718638318745102002001000020030000200372003711102011009910010000100000000710031633199691001002003820038200382003820038
1020420037150000619937251020020010000200100001100707984020018200372003718638318745102002001000020030000200372003711102011009910010000100000000710031633199691001002003820038200382003820038
1020420037150000619937251020020010000200100001100707984120018200372003718638318745102002001000020030000200372003711102011009910010000100000000710031633199691001002003820038200382003820038
1020420037150000619937251020020010000200100001100707984020018200372003718638318745102002001000020030000200372003711102011009910010000100000000710131633199691001002003820038200382003820038
102042003715001288619937251020020010000200100001100707984020018200372003718638318745102002001000020030000200372003711102011009910010000100000000710131633199691001002003820038200382003820038
10204200371500009439937251020020010000200100001100707984020018200372003718638318745102002001000020030000200372003711102011009910010000100000000710131633199691001002003820038200382003820038
102042003715003930619937251020020010000200100001100707984020018200372003718638318745102002001000020030000200372003711102011009910010000100000000710131633199691001002003820038200382003820038
1020420037150000619937251020020010000200100001100707984120018200372003718638318745102002001000020030000200372003711102011009910010000100000000710131643199691001002003820038200382003820038
1020420037150000619937251020020010000200100001100707984020018200372003718638318745102002001000020030000200372003711102011009910010000100000000710131633199691001002003820038200382003820038

1000 unrolls and 10 iterations

Result (median cycles for code): 2.0037

retire (01)cycle (02)031e3f4e51inst 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 integer (97)inst neon or fp (9a)9fa6a8a9acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)gpr retires (ef)f5f6f7f8fd
1002420037150061993725100202010000201000011070798412001820037200371866031876710020201000020300002003720037111002110910100001000000000640216221996910102003820038200382003820038
10024200371506961993725100202010000201000011070798402001820037200371866031876710020201000020300002003720037111002110910100001000000000640216221996910102003820038200382003820038
100242003715037561993725100202010000201000011070798402001820037200371866031876710020201000020300002003720037111002110910100001000000000640216221996910102003820038200382003820038
1002420037150061993725100202010000201000011070798412001820037200371866031876710020201000020300002003720037111002110910100001000000000640216221996910102003820038200382003820038
1002420037150061993725100202010000201000011070798412001820037200371866031876710020201000020300002003720037111002110910100001000000000640216221996910102003820038200382003820038
1002420037150061993725100202010000201000011070798412001820037200371866031876710020201000020300002003720037111002110910100001000000000640216231996910102003820038200382003820038
1002420037150061993725100202010000201000011070798412001820037200371866031876710020201000020300002003720037111002110910100001000000000640216221996910102003820038200382003820038
1002420037150061993725100202010000201000011070798412001820037200371866031876710020201000020300002003720037111002110910100001000000000640216221996910102003820038200382003820038
10024200371500726993725100202010000201000011070798412001820037200371866031876710020201000020300002003720037111002110910100001000000000640216221996910102003820038200382003820038
1002420037150061993725100202010000201000011070798402001820037200371866031876710020201000020300002003720037111002110910100001000000000640216221996910102003820038200382003820038

Test 5: throughput

Count: 8

Code:

  ands xzr, xzr, xzr
  fccmpe s0, s1, #0, lt
  ands xzr, xzr, xzr
  fccmpe s0, s1, #0, lt
  ands xzr, xzr, xzr
  fccmpe s0, s1, #0, lt
  ands xzr, xzr, xzr
  fccmpe s0, s1, #0, lt
  ands xzr, xzr, xzr
  fccmpe s0, s1, #0, lt
  ands xzr, xzr, xzr
  fccmpe s0, s1, #0, lt
  ands xzr, xzr, xzr
  fccmpe s0, s1, #0, lt
  ands xzr, xzr, xzr
  fccmpe s0, s1, #0, lt
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 1.0005

retire (01)cycle (02)030718191e1f3a3f51inst 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)73scheduler 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)9fa8acc2c5branch mispredict (cb)cfd5d6ddinst fetch restart (de)e0? int output thing (e9)gpr retires (ef)f5f6f7f8fd
160204800415990000007172516010080100800008010080000400500880000180022800418004159964035999916010080200800002002400008004180041111602011009910080100800008010000000101102162280035800001008004280042800428004280042
16020480041599000000522516010080100800008010080000400500880000180059800418004159964035999916010080200800002002400008004180041111602011009910080100800008010000000101102162280035800001008004280042800428004280042
16020480041599000000522516010080100800008010080000400500880000180022800418004159964035999916010080200800002002400008004180041111602011009910080100800008010000000101102162280035800001008004280042800428004280042
16020480041599000000522516010080100800008010080000400500880000180022800418004159964035999916010080200800002002400008004180041111602011009910080100800008010000000101102162280035800001008004280042800428004280042
16020480041599000600522516010080100800008010080000400500880000180022800418004159964035999916010080200800002002400008004180041111602011009910080100800008010000000101102162280035800001008004280042800428004280042
16020480041599000000732516010080100800008010080000400500880000180022800418004159964035999916010080200800002002400008004180078111602011009910080100800008010000000101102162280035800001008004280042800428004280042
16020480041600000900522516010080100800008010080000400500880000180022800418004159964035999916010080200800002002400008004180041111602011009910080100800008010000000101102165280115800001008024580042800928009280093
1602048014359900000010202516010080100800008010080000400500880000180022800418004159964035999916010080200800002002400008004180041111602011009910080100800008010000000101102162280035800001008004280042800428004280042
16020480041600011367040942516010080100800008010080000400500880000180022800418004159964035999916010080200800002002400008004180041111602011009910080100800008010000000101102162280035800001008004280042800428004280042
16020480041600000000522516010080100800008010080000400500880000180022800418004159964035999916010080200800002002400008004180041111602011009910080100800008010000001101102162280035800001008004280042800428004280042

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 1.0005

retire (01)cycle (02)03191e3a3f51inst 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)73scheduler 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)9fa1a8accdcfd0d2icache miss (d3)d5d6d9dadbddinst fetch restart (de)e0? int output thing (e9)eaebgpr retires (ef)f5f6f7f8fd
1600248004260000057251600108001080000800108000040005088000011800228004180041600200360020160010800208000020240000800418004111160021109108001080000800100000100223113817211332380036800002015108004280042800428004280042
16002480078599000572516001080010800008001080000400050880000118002280041800415998601260020160010800208000020240000800418004111160021109108001080000800100130100223112017211362380036800002015108004280042800428004280042
1600248004160000057251600108001080000800108000040005088000011800228004180041600090360020160010800208000020240000800418004111160021109108001080000800100000100223113517211372180036800002015108004280042800428004280042
160024800416000001499251600108001080000800108000040005088000011800228004180041599860360020160010800208000020240000800418004111160021109108001080000800100000100223113517211473680036800002015108004280042800428004280042
1600248004160000057251600108001080000800108000040005088000011800228004180041599860360020160010800208000020240000800418004111160021109108001080000800100000100223112117211333780036800002015108009280042800428009480042
1600248004160000057251600108001080000800108000040005088000011800228004180041599860360020160010800208000020240000800418004111160021109108001080000800100000100223112017211363680036800002015108004280042800428004280042
1600248004159900080251600108001080000800108000040005088000011800228004180041599860360020160010800208000020240000800418004111160021109108001080000800100000100223112117211353580036800002015108004280042800428004280042
16002480041599000212251600108001080000800108000040005088000011800618004180041599860360020160010800208000020240000800418004111160021109108001080000800100000100223113517211264080036800002015108004280042800428004280042
16002480041599000437251600108001080000800108000040005088000011800228004180041599860360020160010800208000020240000800418004111160021109108001080000800100100100223113717211342180036800002015108004280042800428004280042
16002480041599000722251600108001080000800108000040005088000011800228004180041599860360020160010800208000020240000800418004111160021109108001080000800100000100223113517211293480036800002015108004280042800428004280042