Apple Microarchitecture Research by Dougall Johnson

M1/A14 P-core (Firestorm): Overview | Base Instructions | SIMD and FP Instructions
M1/A14 E-core (Icestorm):  Overview | Base Instructions | SIMD and FP Instructions

CSET (32-bit)

Test 1: uops

Code:

  cset w0, hi
  mov x0, 1

(no loop instructions)

1000 unrolls and 1 iteration

Retires: 1.000

Issues: 1.000

Integer unit issues: 1.001

Load/store unit issues: 0.000

SIMD/FP unit issues: 0.000

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
1004524100110011000300010001000100010011000
1004370100110011000300010001000100010011000
1004364100110011000300010001000100010011000
1004364100110011000300010001000100010011000
1004364100110011000300010001000100010011000
1004364100110011000300010001000100010011000
1004364100110011000300010001000100010011000
1004364100110011000300010001000100010011000
1004364100110011000300010001000100010011000
1004364100110011000300010001000100010011000

Test 2: Latency 1->2

Chain cycles: 1

Code:

  cset w0, hi
  tst x0, 1
  mov x0, 1

(non-fused SUB/CBNZ loop)

100 unrolls and 100 iterations

Result (median cycles for code, minus 1 chain cycle): 1.0030

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
20204200302020120201202085192672020820216202162010110100
20204200302020120201202085194482020820216202162010110100
20204200302020120201202085194482020820216202162010110100
20204200302020120201202085194482020820216202162010110100
20204200302020120201202085194482020820216202162010110100
20204200302020120201202085194482020820216202162010110100
20205200602021520215202515194482020820216202162010110100
20204200302020120201202085194482020820216202162010110100
20204200302020120201202085194482020820216202162010110100
20204200302020120201202085194482020820216202162010110100

1000 unrolls and 10 iterations

Result (median cycles for code, minus 1 chain cycle): 1.0030

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
20024200302002120021200295194552002920036200202001110010
20024200302002120021200205195912002020020200202001110010
20024200302002120021200205195912002020020200202001110010
20024200302002120021200205195912002020020200202001110010
20024200302002120021200205195912002020020200202001110010
20024200302002120021200205195912002020020200202001110010
20024200302002120021200205195912002020020200202001110010
20024200302002120021200205195912002020020200202001110010
20024200302002120021200205195912002020020200202001110010
20024200302002120021200205195912002020020200202001110010

Test 3: throughput

Count: 8

Code:

  cset w0, hi
  cset w1, hi
  cset w2, hi
  cset w3, hi
  cset w4, hi
  cset w5, hi
  cset w6, hi
  cset w7, hi
  mov x0, 0
  cmp x0, x0

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.3342

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
80204269028011580115801202921688012080222802218001480100
80204267528011580115801202932368012080221802248001580100
80204267378011580115801202932368012080224802768005880100
80204267378011580115801202932368012080224802248001580100
80204267378011580115801202932368012080224802248001580100
80204267378011580115801202932368012080224802248001580100
80204267378011580115801202932368012080224802248001580100
80204267378011580115801202932368012080224802248001580100
80204267378011580115801202740268017480280802218001580100
80204267378011580115801202932368012080224802248001580100

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.3340

retire uop (01)cycle (02)schedule uop (52)schedule int uop (53)dispatch int uop (56)int uops in schedulers (59)dispatch uop (78)map int uop (7c)map int uop inputs (7f)? int output thing (e9)? int retires (ef)
80024279858003880038800513894378002080020800208001180010
80024270648002180021800203894378002080020800208001180010
80024267418002180021800204267808002080020801098007280010
80024267288002180021800204267808002080020801098007180010
80024267278002180021800204267808002080020800208001180010
80024267248002180021800204267808002080020800208001180010
80024267238002180021800204267808002080020800208001180010
80024267288002180021800204267808002080020800208001180010
80024267238002180021800204267808002080020800208001180010
80024267238002180021800204267808002080020800208001180010