OV - exec - Loop 11

	Loop Id: 11	Module: exec	Source: stream.c:324-324	Coverage: 20.18%

0x4016f4 LDR	D6, [X13, X14,LSL #3]    [2]

0x4016f8 ADD	X15, X14, #1

0x4016fc ADD	X10, X14, #2

0x401700 ADD	X9, X14, #3

0x401704 ADD	X8, X14, #4

0x401708 ADD	X16, X14, #5

0x40170c ADD	X17, X14, #6

0x401710 ADD	X18, X14, #7

0x401714 STR	D6, [X12, X14,LSL #3]    [1]

0x401718 ADD	X14, X14, #8

0x40171c LDR	D7, [X13, X15,LSL #3]    [2]

0x401720 STR	D7, [X12, X15,LSL #3]    [1]

0x401724 LDR	D16, [X13, X10,LSL #3]    [2]

0x401728 STR	D16, [X12, X10,LSL #3]    [1]

0x40172c LDR	D17, [X13, X9,LSL #3]    [2]

0x401730 STR	D17, [X12, X9,LSL #3]    [1]

0x401734 LDR	D18, [X13, X8,LSL #3]    [2]

0x401738 STR	D18, [X12, X8,LSL #3]    [1]

0x40173c LDR	D19, [X13, X16,LSL #3]    [2]

0x401740 STR	D19, [X12, X16,LSL #3]    [1]

0x401744 LDR	D20, [X13, X17,LSL #3]    [2]

0x401748 STR	D20, [X12, X17,LSL #3]    [1]

0x40174c LDR	D21, [X13, X18,LSL #3]    [2]

0x401750 STR	D21, [X12, X18,LSL #3]    [1]

0x401754 CMP	X11, X14

0x401758 B.NE	4016f4

/home/eoseret/Tools/QaaS/qaas_runs/175-491-9774/intel/stream/build/stream/src/stream.c: 324 - 324

--------------------------------------------------------------------------------

324: 	    c[j] = a[j];

Coverage (%)	Name	Source Location	Module
►98.43+	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	pthread_condattr_setpshared		libc.so.6
○	__clone		libc.so.6
►1.57+	GOMP_parallel	libgomp.h:980	libgomp.so.1.0.0
○	main	stream.c:327	exec
○	__libc_init_first		libc.so.6
○	__libc_start_main		libc.so.6
○	_start	stream.c:290	exec

min	med	avg	max

Percentile Index	10	20	30	40	50	60	70	80	90	100
Value

min	med	avg	max

Percentile Index	10	20	30	40	50	60	70	80	90	100
Value

Path /

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	4.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.33
Bottlenecks	P10, P11, P12,
Function	main._omp_fn.4
Source	stream.c:324-324
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	5.33
CQA cycles if no scalar integer	5.33
CQA cycles if FP arith vectorized	5.33
CQA cycles if fully vectorized	1.33
Front-end cycles	3.25
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	2.25
P3 cycles	2.25
P4 cycles	2.25
P5 cycles	2.25
P6 cycles	4.00
P7 cycles	4.00
P8 cycles	0.00
P9 cycles	0.00
P10 cycles	5.33
P11 cycles	5.33
P12 cycles	5.33
P13 cycles	0.00
P14 cycles	0.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	26.00
Nb uops	26.00
Nb loads	NA
Nb stores	8.00
Nb stack references	0.00
FLOP/cycle	0.00
Nb FLOP add-sub	0.00
Nb FLOP mul	0.00
Nb FLOP fma	0.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	24.00
Bytes prefetched	0.00
Bytes loaded	64.00
Bytes stored	64.00
Stride 0	0.00
Stride 1	2.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	0.00
Vectorization ratio load	0.00
Vectorization ratio store	0.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	25.00
Vector-efficiency ratio load	25.00
Vector-efficiency ratio store	25.00
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	NA

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	4.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.33
Bottlenecks	P10, P11, P12,
Function	main._omp_fn.4
Source	stream.c:324-324
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	5.33
CQA cycles if no scalar integer	5.33
CQA cycles if FP arith vectorized	5.33
CQA cycles if fully vectorized	1.33
Front-end cycles	3.25
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	2.25
P3 cycles	2.25
P4 cycles	2.25
P5 cycles	2.25
P6 cycles	4.00
P7 cycles	4.00
P8 cycles	0.00
P9 cycles	0.00
P10 cycles	5.33
P11 cycles	5.33
P12 cycles	5.33
P13 cycles	0.00
P14 cycles	0.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	26.00
Nb uops	26.00
Nb loads	NA
Nb stores	8.00
Nb stack references	0.00
FLOP/cycle	0.00
Nb FLOP add-sub	0.00
Nb FLOP mul	0.00
Nb FLOP fma	0.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	24.00
Bytes prefetched	0.00
Bytes loaded	64.00
Bytes stored	64.00
Stride 0	0.00
Stride 1	2.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	0.00
Vectorization ratio load	0.00
Vectorization ratio store	0.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	25.00
Vector-efficiency ratio load	25.00
Vector-efficiency ratio store	25.00
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	NA

Path /

Average path: Display a virtual path defined by average values of all real paths

Function	main._omp_fn.4
Source file and lines	stream.c:324-324
Module	exec

The loop is defined in /home/eoseret/Tools/QaaS/qaas_runs/175-491-9774/intel/stream/build/stream/src/stream.c:324.

The related source loop is not unrolled or unrolled with no peel/tail loop.

gain
potential
hint
expert

Vectorization

Your loop is not vectorized. 4 data elements could be processed at once in vector registers. By vectorizing your loop, you can lower the cost of an iteration from 5.33 to 1.33 cycles (4.00x speedup).

Details

All VPU instructions are used in scalar version (process only one data element in vector registers). Since your execution units are vector units, only a vectorized loop can use their full power.

Workaround

Try another compiler or update/tune your current one:
- recompile with ftree-vectorize (included in O3) to enable loop vectorization (including FP reductions).
Remove inter-iterations dependences from your loop and make it unit-stride:
- If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly: C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)
- If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA): for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)

No data for this section

Matching between your loop (in the source code) and the binary loop

The binary loop does not contain any FP arithmetical operations. The binary loop is loading 64 bytes. The binary loop is storing 64 bytes.

General properties

nb instructions	26
nb uops	26
loop length	104
used w registers	0
used x registers	11
used b registers	0
used h registers	0
used s registers	0
used d registers	8
used q registers	0
used v registers	0
used z registers	0
nb stack references	0

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	3.25 cycles
front end	3.25 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	0.50	0.50	2.25	2.25	2.25	2.25	4.00	4.00	0.00	0.00	5.33	5.33	5.33	0.00	0.00
cycles	0.50	0.50	2.25	2.25	2.25	2.25	4.00	4.00	0.00	0.00	5.33	5.33	5.33	0.00	0.00

Execution ports to units layout:

P0: BRU
P1: BRU
P2: ALU
P3: ALU
P4: ALU
P5: ALU
P6 (128 bits): VPU, FP store data, ALU, DIV/SQRT
P7 (128 bits): VPU, ALU, FP store data
P8 (128 bits): VPU, ALU, DIV/SQRT
P9 (128 bits): ALU, VPU
P10 (256 bits): store address, load
P11 (256 bits): store address, load
P12 (256 bits): load
P13 (64 bits): store data
P14 (64 bits): store data

Cycles executing div or sqrt instructions	NA
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	3.25
Dispatch	5.33
Data deps.	1.00
Overall L1	5.33

Vectorization ratios

all	0%
load	0%
store	0%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

Vector efficiency ratios

all	25%
load	25%
store	25%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 5.33 cycles. At this rate:

12% of peak load performance is reached (12.00 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
25% of peak store performance is reached (12.00 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the loop is: 4016f4

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P10	P11	P12	Latency	Recip. throughput	Vectorization
LDR D6, [X13, X14,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
ADD X15, X14, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X10, X14, #2	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X9, X14, #3	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X8, X14, #4	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X16, X14, #5	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X17, X14, #6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X18, X14, #7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
STR D6, [X12, X14,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
ADD X14, X14, #8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
LDR D7, [X13, X15,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D7, [X12, X15,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D16, [X13, X10,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D16, [X12, X10,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D17, [X13, X9,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D17, [X12, X9,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D18, [X13, X8,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D18, [X12, X8,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D19, [X13, X16,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D19, [X12, X16,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D20, [X13, X17,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D20, [X12, X17,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D21, [X13, X18,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D21, [X12, X18,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
CMP X11, X14	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.33	N/A
B.NE 4016f4 <main._omp_fn.4+0xe4>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	1	0.50	N/A

Function	main._omp_fn.4
Source file and lines	stream.c:324-324
Module	exec

The loop is defined in /home/eoseret/Tools/QaaS/qaas_runs/175-491-9774/intel/stream/build/stream/src/stream.c:324.

The related source loop is not unrolled or unrolled with no peel/tail loop.

gain
potential
hint
expert

Vectorization

Your loop is not vectorized. 4 data elements could be processed at once in vector registers. By vectorizing your loop, you can lower the cost of an iteration from 5.33 to 1.33 cycles (4.00x speedup).

Details

All VPU instructions are used in scalar version (process only one data element in vector registers). Since your execution units are vector units, only a vectorized loop can use their full power.

Workaround

Try another compiler or update/tune your current one:
- recompile with ftree-vectorize (included in O3) to enable loop vectorization (including FP reductions).
Remove inter-iterations dependences from your loop and make it unit-stride:
- If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly: C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)
- If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA): for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)

No data for this section

Matching between your loop (in the source code) and the binary loop

The binary loop does not contain any FP arithmetical operations. The binary loop is loading 64 bytes. The binary loop is storing 64 bytes.

General properties

nb instructions	26
nb uops	26
loop length	104
used w registers	0
used x registers	11
used b registers	0
used h registers	0
used s registers	0
used d registers	8
used q registers	0
used v registers	0
used z registers	0
nb stack references	0

Front-end

MACRO FUSION NOT POSSIBLE FIT IN UOP CACHE

micro-operation queue	3.25 cycles
front end	3.25 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	0.50	0.50	2.25	2.25	2.25	2.25	4.00	4.00	0.00	0.00	5.33	5.33	5.33	0.00	0.00
cycles	0.50	0.50	2.25	2.25	2.25	2.25	4.00	4.00	0.00	0.00	5.33	5.33	5.33	0.00	0.00

Execution ports to units layout:

P0: BRU
P1: BRU
P2: ALU
P3: ALU
P4: ALU
P5: ALU
P6 (128 bits): VPU, FP store data, ALU, DIV/SQRT
P7 (128 bits): VPU, ALU, FP store data
P8 (128 bits): VPU, ALU, DIV/SQRT
P9 (128 bits): ALU, VPU
P10 (256 bits): store address, load
P11 (256 bits): store address, load
P12 (256 bits): load
P13 (64 bits): store data
P14 (64 bits): store data

Cycles executing div or sqrt instructions	NA
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	3.25
Dispatch	5.33
Data deps.	1.00
Overall L1	5.33

Vectorization ratios

all	0%
load	0%
store	0%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

Vector efficiency ratios

all	25%
load	25%
store	25%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 5.33 cycles. At this rate:

12% of peak load performance is reached (12.00 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
25% of peak store performance is reached (12.00 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the loop is: 4016f4

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P10	P11	P12	Latency	Recip. throughput	Vectorization
LDR D6, [X13, X14,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
ADD X15, X14, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X10, X14, #2	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X9, X14, #3	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X8, X14, #4	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X16, X14, #5	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X17, X14, #6	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
ADD X18, X14, #7	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
STR D6, [X12, X14,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
ADD X14, X14, #8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.25	N/A
LDR D7, [X13, X15,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D7, [X12, X15,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D16, [X13, X10,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D16, [X12, X10,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D17, [X13, X9,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D17, [X12, X9,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D18, [X13, X8,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D18, [X12, X8,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D19, [X13, X16,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D19, [X12, X16,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D20, [X13, X17,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D20, [X12, X17,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
LDR D21, [X13, X18,LSL #3]	1	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	scal (25.0%)
STR D21, [X12, X18,LSL #3]	1	0	0	0	0	0	0	0.50	0.50	0.50	0.50	0	2	0.50	scal (25.0%)
CMP X11, X14	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	1	0.33	N/A
B.NE 4016f4 <main._omp_fn.4+0xe4>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	1	0.50	N/A

Report Configuration

Vectorization

Details

Workaround

Matching between your loop (in the source code) and the binary loop

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code

Vectorization

Details

Workaround

Matching between your loop (in the source code) and the binary loop

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code