Firestorm: SQSHRN2 (2D) measurements

Code:

  sqshrn2 v0.4s, v1.2d, #3

Initialization

  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

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

retire uop (01)	cycle (02)	03	mmu table walk data (08)	09	l2 tlb miss data (0b)	18	19	1e	1f	3f	4e	51	schedule uop (52)	schedule simd uop (54)	dispatch simd uop (57)	ldst uops in schedulers (5b)	69	6d	6e	map stall dispatch (70)	map rewind (75)	map stall (76)	dispatch uop (78)	map simd uop (7e)	map simd uop inputs (81)	82	83	flush restart other nonspec (84)	85	inst all (8c)	inst simd alu (9a)	l1d tlb access (a0)	l1d tlb miss (a1)	ld unit uop (a6)	l1d cache writeback (a8)	a9	ac	c2	cf	d5	map dispatch bubble (d6)	dd	fetch restart (de)	e0	? simd retires (ee)	f5	f6	f7	f8	fd
1004	3037	22	0	0	0	0	0	0	0	103	2548	25	1000	1000	1000	398313	3018	3037	3037	2415	3	2895	1000	1000	2000	3037	3037	1	1	1001	1000	0	0	0	0	0	0	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038
1004	3083	23	0	0	0	0	0	0	0	61	2548	25	1000	1000	1000	398313	3018	3037	3037	2415	3	2895	1000	1000	2000	3037	3037	1	1	1001	1000	0	0	0	0	0	0	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038
1004	3037	23	0	0	0	0	0	0	0	103	2548	25	1000	1000	1000	398313	3018	3037	3037	2415	3	2895	1000	1000	2000	3037	3037	1	1	1001	1000	0	0	0	0	0	0	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038
1004	3037	23	0	0	0	0	0	0	0	61	2548	25	1000	1000	1000	398313	3018	3037	3037	2415	3	2895	1000	1000	2000	3037	3037	1	1	1001	1000	0	0	0	0	0	0	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038
1004	3037	23	0	0	0	0	0	0	0	61	2548	25	1000	1000	1000	398313	3018	3037	3037	2415	3	2895	1000	1000	2000	3037	3037	1	1	1001	1000	0	0	0	0	0	0	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038
1004	3037	23	0	0	0	0	0	0	0	84	2548	25	1000	1000	1000	399670	3018	3037	3037	2415	3	2895	1000	1000	2000	3037	3037	1	1	1001	1000	0	0	0	0	0	0	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038
1004	3037	23	0	0	0	0	0	6	0	61	2548	25	1000	1000	1000	398313	3018	3037	3037	2415	3	2895	1000	1000	2000	3037	3037	1	1	1001	1000	0	0	0	0	0	0	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038
1004	3037	22	0	0	0	0	0	0	0	103	2548	25	1000	1000	1000	398313	3018	3037	3037	2415	3	2895	1000	1000	2000	3037	3037	2	1	1001	1000	0	0	0	0	0	0	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038
1004	3037	23	0	0	0	0	1	0	0	61	2548	25	1000	1000	1000	398313	3018	3037	3037	2415	3	2895	1000	1000	2000	3037	3037	1	1	1001	1000	0	0	0	0	0	0	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038
1004	3037	22	0	0	0	0	1	12	88	82	2548	25	1000	1000	1000	398313	3018	3037	3037	2415	3	2915	1000	1000	2000	3037	3037	1	1	1001	1000	0	0	0	0	0	3	0	73	1	16	1	1	2630	1000	3038	3038	3038	3038	3038

Test 2: Latency 1->1

Code:

  sqshrn2 v0.4s, v1.2d, #3

Initialization

  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0037

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

retire uop (01)	cycle (02)	03	mmu table walk data (08)	18	19	1e	1f	3a	3f	4e	51	schedule uop (52)	schedule int uop (53)	schedule simd uop (54)	dispatch int uop (56)	dispatch simd uop (57)	int uops in schedulers (59)	ldst uops in schedulers (5b)	69	6d	6e	map stall dispatch (70)	map rewind (75)	map stall (76)	dispatch uop (78)	map int uop (7c)	map simd uop (7e)	map int uop inputs (7f)	map simd uop inputs (81)	82	83	flush restart other nonspec (84)	85	inst all (8c)	inst branch (8d)	inst branch taken (90)	inst branch cond (94)	inst int alu (97)	inst simd alu (9a)	9f	l1d tlb access (a0)	l1d tlb miss (a1)	ld unit uop (a6)	st unit uop (a7)	l1d cache writeback (a8)	a9	ac	c2	cf	d5	map dispatch bubble (d6)	dd	fetch restart (de)	e0	? int output thing (e9)	? simd retires (ee)	? int retires (ef)	f5	f6	f7	f8	fd
10204	30037	225	0	0	0	0	0	0	168	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	0	0	0	0	0	0	61	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	0	0	0	0	0	0	474	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30087
10204	30037	225	0	0	0	0	0	0	145	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	0	0	0	0	0	0	61	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	0	0	0	0	0	0	475	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	0	0	0	0	0	0	636	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	0	0	0	0	0	0	506	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	0	0	0	0	0	0	510	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	0	0	0	9	0	0	61	29548	25	10100	100	10000	100	10000	500	4277313	30018	30037	30037	28265	3	28745	10100	200	10000	200	20000	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	0	0	0	710	1	16	1	1	29634	0	10000	100	30038	30038	30038	30038	30038

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0037

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

retire uop (01)	cycle (02)	03	mmu table walk data (08)	09	l2 tlb miss data (0b)	18	19	1e	1f	3f	4e	51	schedule uop (52)	schedule int uop (53)	schedule simd uop (54)	dispatch int uop (56)	dispatch simd uop (57)	int uops in schedulers (59)	ldst uops in schedulers (5b)	60	61	69	6b	6d	6e	map stall dispatch (70)	map rewind (75)	map stall (76)	dispatch uop (78)	map int uop (7c)	map simd uop (7e)	map int uop inputs (7f)	map simd uop inputs (81)	82	83	flush restart other nonspec (84)	85	inst all (8c)	inst branch (8d)	inst branch taken (90)	inst branch cond (94)	inst int alu (97)	inst simd alu (9a)	9f	l1d tlb access (a0)	l1d tlb miss (a1)	ld unit uop (a6)	st unit uop (a7)	l1d cache writeback (a8)	a9	ac	c2	cf	d0	d5	map dispatch bubble (d6)	dd	fetch restart (de)	e0	? int output thing (e9)	? simd retires (ee)	? int retires (ef)	f5	f6	f7	f8	fd
10024	30263	225	1	0	1	1	0	12	0	729	29521	44	10028	13	10024	12	10447	60	4281384	1	0	30162	0	30226	30273	28303	7	28803	10313	20	10326	20	20982	30133	30178	4	1	10021	10	9	10	10	10000	10	0	2	2	0	1	2	2723	2	661	0	2	41	2	2	29774	2	10000	10	30134	30133	30132	30121	30651
10024	30180	227	1	0	0	5	1	1068	88	128	29530	81	10058	11	10008	12	10596	77	4289526	0	0	30162	0	30179	30178	28287	24	28841	10160	20	10161	20	20326	30133	30167	2	1	10021	10	9	10	10	10000	10	0	0	0	0	0	0	2583	2	682	0	3	16	2	2	29707	0	10000	10	30133	30038	30209	30132	30085
10024	30274	227	1	0	1	1	1	267	284	759	29503	44	10028	10	10000	10	10000	55	4278670	0	0	30054	0	30227	30037	28291	12	28804	10459	20	10000	22	22284	30132	30037	2	1	10021	10	9	10	10	10000	10	0	0	0	0	0	0	6923	0	661	0	2	16	2	2	29668	3	10000	10	30086	30038	30038	30131	30180
10024	30084	225	0	0	0	0	0	2835	2990	82	29548	25	10010	10	10000	10	10000	50	4280027	0	0	30090	0	30132	30132	28291	3	28786	10038	22	10492	24	20000	30178	30371	2	1	10021	10	9	10	10	10000	10	0	2	0	1	1	0	0	0	661	0	2	16	5	2	29630	2	10000	10	30038	30038	30038	30133	30038
10024	30037	224	0	0	0	0	0	0	0	103	29548	25	10020	12	10000	10	10000	55	4277313	0	0	30018	0	30037	30037	28287	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	0	0	1	0	0	0	640	0	2	16	2	3	29700	0	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	1	0	2	0	0	145	29548	25	10010	10	10000	12	10000	50	4277313	0	0	30018	0	30130	30083	28287	11	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	2	0	0	0	1	0	3	0	640	0	2	16	2	2	29630	0	10000	10	30038	30038	30038	30038	30085
10024	30037	224	0	0	0	0	0	12	0	103	29548	25	10010	10	10000	10	10000	50	4277313	0	0	30018	0	30037	30037	28287	3	28786	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	0	0	1	0	3	0	640	0	2	16	2	2	29630	0	10000	10	30038	30038	30038	30038	30038
10024	30037	225	1	0	0	0	0	0	0	61	29548	25	10010	10	10000	10	10000	50	4277313	0	0	30018	0	30037	30037	28287	21	28914	11656	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	0	0	0	0	0	0	640	0	2	16	2	2	29630	0	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	0	0	0	0	61	29548	25	10010	10	10000	10	10000	50	4277313	1	0	30018	0	30037	30037	28287	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	0	0	0	0	0	0	640	0	2	16	2	2	29630	0	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	0	0	0	0	61	29548	25	10010	10	10000	10	10000	50	4277313	0	3	30018	0	30037	30037	28287	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	0	0	0	0	0	0	640	3	2	16	2	2	29630	0	10000	10	30038	30038	30038	30038	30038

Test 3: Latency 1->2

Code:

  sqshrn2 v0.4s, v0.2d, #3

Initialization

  movi v0.16b, 1

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0037

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

retire uop (01)	cycle (02)	03	mmu table walk instruction (07)	l2 tlb miss instruction (0a)	1e	3f	4e	51	schedule uop (52)	schedule int uop (53)	schedule simd uop (54)	dispatch int uop (56)	dispatch simd uop (57)	int uops in schedulers (59)	ldst uops in schedulers (5b)	69	6d	6e	map stall dispatch (70)	map rewind (75)	map stall (76)	dispatch uop (78)	map int uop (7c)	map simd uop (7e)	map int uop inputs (7f)	map simd uop inputs (81)	82	83	flush restart other nonspec (84)	85	inst all (8c)	inst branch (8d)	inst branch taken (90)	inst branch cond (94)	inst int alu (97)	inst simd alu (9a)	9f	ld unit uop (a6)	l1d cache writeback (a8)	a9	ac	c2	branch cond mispred nonspec (c5)	branch mispred nonspec (cb)	cd	cf	d5	map dispatch bubble (d6)	dd	fetch restart (de)	e0	? int output thing (e9)	? simd retires (ee)	? int retires (ef)	f5	f6	f7	f8	fd
10204	30037	225	1	1	432	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	6	28740	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	1	1	1	717	1	16	1	1	29648	0	10000	100	30038	30038	30085	30038	30038
10204	30037	225	1	1	294	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	6	28741	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	1	1	1	717	1	16	1	1	29647	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	1	1	498	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	6	28741	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	1	1	1	717	1	16	1	1	29647	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	1	1	489	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	7	28740	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	1	1	1	718	1	16	1	1	29647	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	1	1	498	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	6	28774	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	3	0	0	0	1	1	1	717	1	16	1	1	29648	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	1	1	483	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	6	28740	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	1	1	1	717	1	16	1	1	29648	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	1	1	423	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	6	28740	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	1	1	1	717	1	16	1	1	29648	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	1	1	0	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	7	28740	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	22	0	3	0	1	1	1	717	1	16	1	1	29647	0	10000	100	30038	30038	30038	30038	30038
10204	30037	224	1	1	519	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	6	28740	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	1	1	1	718	1	16	1	1	29648	0	10000	100	30038	30038	30038	30038	30038
10204	30037	225	1	1	465	61	29547	25	10100	100	10000	100	10000	500	4277160	30018	30037	30037	28271	7	28741	10100	200	10008	200	20016	30037	30037	1	1	10201	100	99	100	100	10000	100	0	0	0	0	0	1	1	1	717	1	16	1	1	29648	0	10000	100	30038	30038	30038	30038	30038

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0037

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

retire uop (01)	cycle (02)	03	mmu table walk data (08)	1e	1f	3f	4e	51	schedule uop (52)	schedule int uop (53)	schedule simd uop (54)	dispatch int uop (56)	dispatch simd uop (57)	int uops in schedulers (59)	ldst uops in schedulers (5b)	60	69	6b	6d	6e	map stall dispatch (70)	simd prf full (72)	map rewind (75)	map stall (76)	dispatch uop (78)	map int uop (7c)	map simd uop (7e)	map int uop inputs (7f)	map simd uop inputs (81)	82	83	flush restart other nonspec (84)	85	inst all (8c)	inst branch (8d)	inst branch taken (90)	inst branch cond (94)	inst int alu (97)	inst simd alu (9a)	9f	ld unit uop (a6)	l1d cache writeback (a8)	cf	d5	map dispatch bubble (d6)	dd	fetch restart (de)	e0	? simd retires (ee)	? int retires (ef)	f5	f6	f7	f8	fd
10024	30037	224	0	0	0	992	29547	25	10010	10	10000	10	10000	50	4277160	1	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	61	29547	25	10010	10	10000	10	10000	50	4277160	0	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	61	29547	25	10010	10	10000	10	10000	50	4277160	1	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	61	29547	25	10010	10	10000	10	10000	50	4277160	1	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	61	29547	25	10010	10	10000	10	10000	50	4277160	1	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	61	29547	25	10010	10	10000	10	10000	50	4277160	1	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	61	29547	25	10010	10	10000	10	10000	50	4277160	1	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	61	29547	25	10010	10	10000	10	10000	50	4277160	1	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30085	30038	30038	30038	30038
10024	30037	225	0	0	0	61	29547	25	10010	10	10000	10	10000	50	4277160	1	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30038	30038	30038	30038	30038
10024	30037	225	0	0	0	61	29547	25	10010	10	10000	10	10000	50	4277160	0	30018	0	30037	30037	28286	0	3	28767	10010	20	10000	20	20000	30037	30037	1	1	10021	10	9	10	10	10000	10	0	0	640	2	16	2	2	29629	10000	10	30038	30038	30038	30038	30038

Test 4: throughput

Count: 8

Code:

  movi v0.16b, 0
  sqshrn2 v0.4s, v8.2d, #3
  movi v1.16b, 0
  sqshrn2 v1.4s, v8.2d, #3
  movi v2.16b, 0
  sqshrn2 v2.4s, v8.2d, #3
  movi v3.16b, 0
  sqshrn2 v3.4s, v8.2d, #3
  movi v4.16b, 0
  sqshrn2 v4.4s, v8.2d, #3
  movi v5.16b, 0
  sqshrn2 v5.4s, v8.2d, #3
  movi v6.16b, 0
  sqshrn2 v6.4s, v8.2d, #3
  movi v7.16b, 0
  sqshrn2 v7.4s, v8.2d, #3

Initialization

  movi v8.16b, 9

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.2510

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

retire uop (01)	cycle (02)	03	1e	3f	51	schedule uop (52)	schedule int uop (53)	schedule simd uop (54)	dispatch int uop (56)	dispatch simd uop (57)	int uops in schedulers (59)	ldst uops in schedulers (5b)	60	69	6d	6e	map rewind (75)	map stall (76)	dispatch uop (78)	map int uop (7c)	map simd uop (7e)	map int uop inputs (7f)	map simd uop inputs (81)	82	83	flush restart other nonspec (84)	85	inst all (8c)	inst branch (8d)	inst branch taken (90)	inst branch cond (94)	inst int alu (97)	inst simd alu (9a)	9f	l1d cache writeback (a8)	ac	branch cond mispred nonspec (c5)	branch mispred nonspec (cb)	cd	cf	d5	map dispatch bubble (d6)	dd	fetch restart (de)	e0	? simd retires (ee)	? int retires (ef)	f5	f6	f7	f8	fd
160204	20088	150	378	29	25	80116	100	80016	100	80028	500	640196	1	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066
160204	20065	156	0	50	25	80116	100	80016	100	80028	500	640196	0	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066
160204	20065	151	0	29	25	80116	100	80016	100	80028	500	640196	0	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066
160204	20065	150	0	29	25	80116	100	80016	100	80028	500	640196	0	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066
160204	20065	150	0	504	25	80116	100	80016	100	80028	500	640196	0	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066
160204	20065	150	0	29	25	80116	100	80016	100	80028	500	640196	0	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066
160204	20065	150	0	29	25	80116	100	80016	100	80028	500	640196	0	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066
160204	20065	150	339	29	25	80116	100	80016	100	80028	500	640196	0	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066
160204	20065	150	21	29	25	80116	100	80016	100	80028	500	640196	0	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066
160204	20065	150	0	29	25	80116	100	80016	100	80028	500	640196	0	20045	20065	20065	6	12	80128	200	80028	200	160056	20065	20065	1	1	160201	100	99	100	100	160000	100	0	0	1	1	1	10119	0	16	0	0	20062	160000	100	20066	20066	20066	20066	20066

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.2506

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

retire uop (01)	cycle (02)	03	mmu table walk data (08)	09	l2 tlb miss data (0b)	18	19	1e	1f	3a	3f	50	51	schedule uop (52)	schedule int uop (53)	schedule simd uop (54)	dispatch int uop (56)	dispatch simd uop (57)	int uops in schedulers (59)	ldst uops in schedulers (5b)	5f	60	61	69	6d	6e	map rewind (75)	map stall (76)	dispatch uop (78)	map int uop (7c)	map simd uop (7e)	map int uop inputs (7f)	map simd uop inputs (81)	82	83	flush restart other nonspec (84)	85	inst all (8c)	inst branch (8d)	inst branch taken (90)	inst branch cond (94)	inst int alu (97)	inst simd alu (9a)	9f	l1d tlb access (a0)	l1d tlb miss (a1)	ld unit uop (a6)	st unit uop (a7)	a9	ac	c2	cf	d0	d2	l1i cache miss demand (d3)	d5	map dispatch bubble (d6)	d9	da	db	dd	fetch restart (de)	e0	? int output thing (e9)	ea	ec	? simd retires (ee)	? int retires (ef)	f5	f6	f7	f8	fd
160024	20074	150	0	0	0	0	0	3	0	0	45	0	25	80010	10	80000	10	80000	50	640000	1	1	5	20027	20050	20050	3	22	80010	20	80000	20	160000	20046	20046	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10041	8	3	1	21	20	2	1	1	18	19	20043	0	15	0	160000	10	20047	20047	20051	20047	20047
160024	20046	151	0	1	0	0	0	486	0	1	51	0	25	80010	10	80000	10	80000	50	640000	1	1	5	20027	20046	20046	3	22	80010	20	80000	20	160000	20046	20050	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10041	8	2	1	18	20	2	1	1	18	19	20043	0	15	0	160000	10	20047	20047	20047	20047	20047
160024	20046	150	0	0	0	0	0	486	0	0	45	0	25	80010	10	80000	10	80000	50	640000	1	1	5	20098	20046	20046	3	22	80010	20	80000	20	160000	20046	20046	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10041	3	2	1	18	20	2	1	1	19	19	20043	0	15	0	160000	10	20047	20047	20047	20047	20047
160024	20046	150	0	0	0	0	0	0	0	0	45	0	25	80010	10	80000	10	80000	50	640000	1	0	0	20027	20046	20046	3	22	80010	20	80000	20	160000	20046	20046	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	2473	0	10191	3	2	1	26	215	2	1	1	22	20	20376	0	15	0	160000	10	20186	20504	20453	20454	20616
160024	20368	153	1	0	1	6	6	801	528	0	66	0	168	80115	10	80632	10	80000	50	645032	1	1	5	20027	20287	20529	27	150	80327	20	80522	20	161262	20046	20526	3	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10046	8	2	1	16	20	2	1	1	16	18	20043	0	15	0	160000	10	20047	20047	20047	20047	20047
160024	20046	150	0	0	0	0	0	0	0	0	45	0	25	80010	10	80000	10	80000	50	640000	1	0	5	20027	20050	20046	3	22	80432	20	80000	20	160000	20046	20046	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10040	8	2	1	19	20	2	1	1	19	18	20043	0	15	0	160000	10	20047	20047	20047	20047	20047
160024	20046	150	0	0	0	0	0	0	0	0	45	0	25	80010	10	80000	10	80000	50	640000	1	1	5	20027	20046	20046	3	22	80010	20	80000	20	160000	20046	20046	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10044	8	2	1	18	20	2	1	1	20	19	20043	0	15	0	160000	10	20047	20047	20047	20047	20047
160024	20046	151	0	0	0	0	0	333	0	0	45	0	25	80010	10	80000	10	80000	50	640000	1	0	5	20027	20046	20046	3	22	80010	20	80000	20	160000	20046	20046	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10041	8	2	1	20	20	2	1	2	18	18	20043	0	30	0	160000	10	20051	20047	20047	20047	20047
160024	20050	150	0	0	0	0	0	0	0	0	237	0	25	80010	10	80000	10	80000	50	640000	1	1	5	20027	20046	20046	3	22	80010	20	80000	20	160000	20046	20046	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10042	3	1	1	19	20	2	1	2	19	19	20047	0	15	0	160000	10	20047	20051	20047	20047	20047
160024	20046	150	0	0	0	0	0	0	0	0	45	0	25	80010	10	80000	10	80000	50	640000	1	0	0	20027	20046	20046	3	22	80010	20	80000	20	160000	20046	20046	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10040	8	2	2	17	20	2	1	1	19	18	20043	0	15	0	160000	10	20047	20047	20047	20047	20047

Test 5: throughput

Count: 16

Code:

  sqshrn2 v0.4s, v16.2d, #3
  sqshrn2 v1.4s, v16.2d, #3
  sqshrn2 v2.4s, v16.2d, #3
  sqshrn2 v3.4s, v16.2d, #3
  sqshrn2 v4.4s, v16.2d, #3
  sqshrn2 v5.4s, v16.2d, #3
  sqshrn2 v6.4s, v16.2d, #3
  sqshrn2 v7.4s, v16.2d, #3
  sqshrn2 v8.4s, v16.2d, #3
  sqshrn2 v9.4s, v16.2d, #3
  sqshrn2 v10.4s, v16.2d, #3
  sqshrn2 v11.4s, v16.2d, #3
  sqshrn2 v12.4s, v16.2d, #3
  sqshrn2 v13.4s, v16.2d, #3
  sqshrn2 v14.4s, v16.2d, #3
  sqshrn2 v15.4s, v16.2d, #3

Initialization

  movi v16.16b, 17

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.2502

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

retire uop (01)	cycle (02)	03	mmu table walk data (08)	09	18	19	1e	1f	3f	51	schedule uop (52)	schedule int uop (53)	schedule simd uop (54)	dispatch int uop (56)	dispatch simd uop (57)	int uops in schedulers (59)	ldst uops in schedulers (5b)	69	6d	6e	map stall dispatch (70)	map rewind (75)	map stall (76)	dispatch uop (78)	map int uop (7c)	map simd uop (7e)	map int uop inputs (7f)	map simd uop inputs (81)	82	83	flush restart other nonspec (84)	85	inst all (8c)	inst branch (8d)	inst branch taken (90)	inst branch cond (94)	inst int alu (97)	inst simd alu (9a)	9f	l1d tlb access (a0)	ld unit uop (a6)	l1d cache writeback (a8)	a9	ac	c2	branch cond mispred nonspec (c5)	branch mispred nonspec (cb)	cd	cf	d5	map dispatch bubble (d6)	dd	fetch restart (de)	e0	? int output thing (e9)	? simd retires (ee)	? int retires (ef)	f5	f6	f7	f8	fd
160204	40059	300	0	0	0	0	24	0	30	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040
160204	40039	300	0	0	0	0	0	0	30	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040
160204	40039	300	0	0	0	0	0	0	30	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040
160204	40039	301	0	0	0	0	0	0	695	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040
160204	40039	300	0	0	0	0	0	0	30	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040
160204	40039	300	0	0	0	0	0	0	30	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040
160204	40039	300	0	0	0	0	0	0	30	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040
160204	40039	300	0	0	0	0	27	0	30	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040
160204	40039	299	0	0	0	0	0	0	30	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040
160204	40039	300	0	0	0	0	0	0	30	25	160108	100	160008	100	160020	500	1280132	40020	40039	40039	19977	6	19990	160120	200	160032	200	320064	40039	40039	1	1	160201	100	99	100	100	160000	100	0	0	0	0	0	0	1	1	1	10118	0	16	0	0	40036	0	160000	100	40040	40040	40040	40040	40040

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.2502

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

retire uop (01)	cycle (02)	03	l1i tlb fill (04)	mmu table walk data (08)	09	1e	3f	51	schedule uop (52)	schedule int uop (53)	schedule simd uop (54)	dispatch int uop (56)	dispatch simd uop (57)	int uops in schedulers (59)	ldst uops in schedulers (5b)	5f	60	69	6d	6e	map stall dispatch (70)	map rewind (75)	map stall (76)	dispatch uop (78)	map int uop (7c)	map simd uop (7e)	map int uop inputs (7f)	map simd uop inputs (81)	82	83	flush restart other nonspec (84)	85	inst all (8c)	inst branch (8d)	inst branch taken (90)	inst branch cond (94)	inst int alu (97)	inst simd alu (9a)	9f	l1d tlb access (a0)	l1d tlb miss (a1)	l1d cache writeback (a8)	ac	c2	branch cond mispred nonspec (c5)	branch mispred nonspec (cb)	cf	d0	d2	l1i cache miss demand (d3)	d5	map dispatch bubble (d6)	d9	da	db	dd	fetch restart (de)	e0	ea	eb	? simd retires (ee)	? int retires (ef)	f5	f6	f7	f8	fd
160024	40050	300	0	1	0	6	52	81	160010	10	160000	10	160000	50	1280000	0	1	40020	40039	40039	19996	3	20019	160010	20	160000	20	320000	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10024	6	2	2	9	16	4	2	2	4	2	40036	15	5	160000	10	40040	40040	40040	40040	40040
160024	40039	300	0	0	0	6	330	25	160010	10	160000	10	160000	50	1280000	0	1	40020	40039	40039	19996	3	20019	160010	20	160000	20	320000	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10024	3	2	2	2	16	4	2	2	2	4	40036	30	10	160000	10	40040	40040	40040	40040	40040
160024	40039	300	0	0	0	0	52	25	160010	10	160000	10	160000	50	1280000	0	0	40020	40039	40245	19996	3	20019	160010	20	160000	20	320000	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10022	3	1	1	4	16	2	1	1	4	2	40036	30	10	160000	10	40040	40040	40040	40040	40040
160024	40039	299	0	0	0	0	46	25	160010	10	160000	10	160000	50	1280000	0	1	40020	40039	40039	19996	3	20019	160010	20	160000	20	320000	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10022	3	1	1	7	16	2	2	1	8	8	40036	15	5	160000	10	40040	40040	40040	40040	40040
160024	40039	300	0	0	0	0	46	25	160010	10	160000	10	160000	50	1280000	0	1	40020	40039	40039	19996	3	20019	160010	20	160000	20	320000	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10022	3	1	1	2	16	2	1	2	4	8	40036	15	10	160000	10	40040	40040	40040	40040	40088
160024	40039	300	0	0	0	0	46	25	160010	10	160000	10	160000	50	1280000	1	1	40020	40039	40039	19996	3	20019	160010	20	160000	20	320000	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10022	6	2	2	5	16	2	1	1	4	5	40036	15	5	160000	10	40040	40040	40040	40040	40040
160024	40039	300	0	0	0	0	46	25	160010	10	160000	10	160000	50	1280000	1	1	40020	40039	40039	19996	3	20019	160010	20	160000	20	320000	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10022	3	1	1	2	16	2	1	1	4	2	40036	15	5	160000	10	40040	40040	40040	40040	40040
160024	40039	299	0	0	0	0	711	25	160010	10	160000	10	160000	50	1280000	1	1	40020	40039	40246	19996	3	20019	160010	20	160000	20	320616	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10024	3	1	1	2	16	4	1	1	2	4	40036	15	10	160000	10	40040	40040	40040	40040	40040
160024	40039	300	0	0	0	0	46	25	160010	10	160000	10	160000	50	1280000	1	1	40020	40039	40039	19996	3	20019	160010	20	160000	20	320000	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10022	3	1	1	4	16	2	1	1	4	4	40036	15	10	160000	10	40040	40040	40040	40040	40040
160024	40039	300	0	0	0	0	46	25	160010	10	160000	10	160000	50	1280000	1	1	40020	40039	40039	19996	3	20019	160010	20	160000	20	320000	40039	40039	1	1	160021	10	9	10	10	160000	10	0	0	0	0	0	0	0	10022	6	1	1	3	16	4	2	1	2	4	40036	15	10	160000	10	40040	40040	40040	40040	40040

SQSHRN2 (2D)

Test 1: uops

1000 unrolls and 1 iteration

Test 2: Latency 1->1

100 unrolls and 100 iterations

1000 unrolls and 10 iterations

Test 3: Latency 1->2

100 unrolls and 100 iterations

1000 unrolls and 10 iterations

Test 4: throughput

100 unrolls and 100 iterations

1000 unrolls and 10 iterations

Test 5: throughput

100 unrolls and 100 iterations

1000 unrolls and 10 iterations