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

REV16 (32-bit)

Test 1: uops

Code:

  rev16 w0, w0

Initialization

  mov x0, 1
  mov x1, 2

(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

Recorded non-zero counters (first 10 of 64 runs)

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)
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000
1004	1030	1001	1001	1000	25603	1000	1000	1000	1001	1000

Test 2: Latency 1->2

Code:

  rev16 w0, w0

Initialization

  mov x0, 1
  mov x1, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 1.0030

Recorded non-zero counters (first 10 of 64 runs)

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)
10204	10030	10101	10101	10108	259416	10107	10214	10214	10001	10100
10204	10030	10101	10101	10107	259539	10107	10212	10212	10001	10100
10204	10030	10101	10101	10107	259539	10107	10212	10212	10001	10100
10204	10030	10101	10101	10107	259539	10107	10212	10212	10001	10100
10204	10030	10101	10101	10107	259539	10107	10212	10212	10001	10100
10204	10030	10101	10101	10107	259539	10107	10212	10212	10001	10100
10204	10030	10101	10101	10107	259539	10107	10212	10212	10001	10100
10204	10030	10101	10101	10107	259539	10107	10212	10212	10001	10100
10204	10030	10101	10101	10107	259539	10107	10212	10212	10001	10100
10204	10030	10101	10101	10107	259539	10107	10212	10212	10001	10100

1000 unrolls and 10 iterations

Result (median cycles for code): 1.0030

Recorded non-zero counters (first 10 of 64 runs)

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)
10024	10030	10021	10021	10028	259628	10028	10036	10020	10011	10010
10024	10030	10021	10021	10020	259591	10020	10020	10020	10011	10010
10024	10030	10021	10021	10020	259591	10020	10020	10020	10011	10010
10024	10030	10021	10021	10020	259591	10020	10020	10020	10011	10010
10024	10030	10021	10021	10020	259591	10020	10020	10020	10011	10010
10024	10030	10021	10021	10020	259591	10020	10020	10020	10011	10010
10024	10030	10021	10021	10020	259591	10020	10020	10020	10011	10010
10024	10030	10021	10021	10020	259591	10020	10020	10020	10011	10010
10024	10030	10021	10021	10020	259591	10020	10020	10020	10011	10010
10024	10030	10021	10021	10020	259591	10020	10020	10020	10011	10010

Test 3: throughput

Count: 8

Code:

  rev16 w0, w8
  rev16 w1, w8
  rev16 w2, w8
  rev16 w3, w8
  rev16 w4, w8
  rev16 w5, w8
  rev16 w6, w8
  rev16 w7, w8

Initialization

  mov x8, 9

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

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

Recorded non-zero counters (first 10 of 64 runs)

retire uop (01)	cycle (02)	schedule uop (52)	schedule int uop (53)	schedule simd uop (54)	schedule ldst uop (55)	dispatch int uop (56)	dispatch simd uop (57)	int uops in schedulers (59)	dispatch uop (78)	map int uop (7c)	map int uop inputs (7f)	? int output thing (e9)	? int retires (ef)
80204	26867	80115	80115	0	0	80120	0	240360	80120	80224	80224	80015	80100
80204	26755	80115	80115	0	0	80120	0	240504	80171	80276	80224	80015	80100
80204	26737	80115	80115	0	0	80120	0	240360	80120	80224	80224	80015	80100
80204	26737	80115	80115	0	0	80120	0	240360	80120	80224	80224	80015	80100
80204	26737	80115	80115	0	0	80120	0	240360	80120	80224	80224	80015	80100
80204	26737	80115	80115	0	0	80120	0	240360	80120	80222	80222	80014	80100
80204	26737	80115	80115	0	0	80120	0	240360	80120	80224	80224	80015	80100
80204	26751	80115	80115	0	0	80120	0	240507	80172	80277	80275	80058	80100
100445	68010	100384	93318	53	7013	93174	51	240507	80172	80275	80224	80015	80100
80204	26737	80115	80115	0	0	80120	0	240360	80120	80224	80224	80015	80100

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.3339

Recorded non-zero counters (first 10 of 64 runs)

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)
80024	27944	80037	80037	80042	277404	80020	80020	80020	80011	80010
80024	26769	80021	80021	80020	277404	80020	80020	80020	80011	80010
80024	26713	80021	80021	80020	277404	80020	80020	80020	80011	80010
80024	26713	80021	80021	80020	277404	80020	80020	80020	80011	80010
80024	26713	80021	80021	80020	277404	80020	80020	80020	80011	80010
80024	26713	80021	80021	80020	277404	80020	80020	80020	80011	80010
80024	26713	80021	80021	80020	277404	80020	80020	80020	80011	80010
80024	26713	80021	80021	80020	277404	80020	80020	80020	80011	80010
80024	26713	80021	80021	80020	277404	80020	80020	80020	80011	80010
80024	26713	80021	80021	80020	277404	80020	80020	80101	80070	80010