OV - exec - Loop 185

0x41c5e0 MOVPRFX	Z20, Z0

0x41c5e4 SDIV	Z20.D, P0/M, Z20.D, Z2.D

0x41c5e8 ADD	Z21.D, Z3.D, Z20.D

0x41c5ec MOVPRFX	Z22, Z0

0x41c5f0 MLS	Z22.D, P0/M, Z20.D, Z2.D

0x41c5f4 ADD	Z20.D, Z6.D, Z20.D

0x41c5f8 SUBS	X12, X12, X10

0x41c5fc ADD	Z23.D, Z4.D, Z22.D

0x41c600 MOVPRFX	Z24, Z23

0x41c604 SXTW	Z24.D, P0/M, Z23.D

0x41c608 ADD	Z22.D, Z17.D, Z22.D

0x41c60c ADD	Z0.D, Z0.D, Z1.D

0x41c610 SXTW	Z20.D, P0/M, Z20.D

0x41c614 SXTW	Z21.D, P0/M, Z21.D

0x41c618 MOVPRFX	Z25, Z24

0x41c61c MLA	Z25.D, P0/M, Z5.D, Z21.D

0x41c620 LD1D	{Z25.D}, P0/Z, [X24, Z25.D,LSL #3]    [3]

0x41c624 MOVPRFX	Z26, Z18

0x41c628 MUL	Z26.D, P0/M, Z26.D, Z21.D

0x41c62c ADR	Z22.D, [Z26, Z22.D,SXTW]    [4]

0x41c630 MAD	Z20.D, P0/M, Z7.D, Z24.D

0x41c634 LD1D	{Z20.D}, P0/Z, [X26, Z20.D,LSL #3]    [2]

0x41c638 FADD	Z20.D, Z20.D, Z25.D

0x41c63c MOVPRFX	Z25, Z24

0x41c640 MLA	Z25.D, P0/M, Z7.D, Z21.D

0x41c644 LD1D	{Z25.D}, P0/Z, [X26, Z25.D,LSL #3]    [8]

0x41c648 FSUB	Z20.D, Z20.D, Z25.D

0x41c64c MOVPRFX	Z25, Z24

0x41c650 MLA	Z25.D, P0/M, Z16.D, Z21.D

0x41c654 MAD	Z21.D, P0/M, Z19.D, Z24.D

0x41c658 ST1D	{Z20.D}, P0, [X22, Z25.D,LSL #3]    [9]

0x41c65c LD1D	{Z22.D}, P0/Z, [X21, Z22.D,LSL #3]    [5]

0x41c660 FADD	Z20.D, Z22.D, Z20.D

0x41c664 ADR	Z22.D, [Z26, Z23.D,SXTW]    [6]

0x41c668 LD1D	{Z22.D}, P0/Z, [X21, Z22.D,LSL #3]    [1]

0x41c66c FSUB	Z20.D, Z20.D, Z22.D

0x41c670 ST1D	{Z20.D}, P0, [X28, Z21.D,LSL #3]    [7]

0x41c674 B.NE	41c5e0

/home/hbollore/qaas-runs/170-290-5445/intel/CloverLeafCXX/build/CloverLeafCXX/src/omp/advec_mom.cpp: 45 - 48

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

45:     for (int j = (y_min - 2 + 1); j < (y_max + 2 + 2); j++) {

46:       for (int i = (x_min - 2 + 1); i < (x_max + 2 + 2); i++) {

47:         post_vol(i, j) = volume(i, j) + vol_flux_y(i + 0, j + 1) - vol_flux_y(i, j);

48:         pre_vol(i, j) = post_vol(i, j) + vol_flux_x(i + 1, j + 0) - vol_flux_x(i, j);

/home/hbollore/qaas-runs/170-290-5445/intel/CloverLeafCXX/build/CloverLeafCXX/src/omp/context.h: 69 - 69

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

69:   T &operator()(size_t i, size_t j) const { return data[i + j * sizeX]; }

Coverage (%)	Name	Source Location	Module
○100.00	__kmp_invoke_microtask		libomp.so

Path /

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	1.05
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	3.05
Bottlenecks	P6, P7,
Function	.omp_outlined.#0x41c480
Source	advec_mom.cpp:45-48,context.h:69-69
Source loop unroll info	unrolled by 4
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	4
CQA cycles	14.50
CQA cycles if no scalar integer	14.50
CQA cycles if FP arith vectorized	14.50
CQA cycles if fully vectorized	13.75
Front-end cycles	4.75
DIV/SQRT cycles	0.50
P0 cycles	0.50
P1 cycles	0.25
P2 cycles	0.25
P3 cycles	0.25
P4 cycles	0.25
P5 cycles	14.50
P6 cycles	14.50
P7 cycles	2.00
P8 cycles	2.00
P9 cycles	3.50
P10 cycles	3.50
P11 cycles	0.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	1.00 - 0.50
Inter-iter dependencies cycles	2
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	38.00
Nb uops	38.00
Nb loads	NA
Nb stores	2.00
Nb stack references	0.00
FLOP/cycle	1.10
Nb FLOP add-sub	16.00
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	15.45
Bytes prefetched	0.00
Bytes loaded	160.00
Bytes stored	64.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	72.97
Vectorization ratio load	71.43
Vectorization ratio store	100.00
Vectorization ratio mul	100.00
Vectorization ratio add_sub	81.25
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	100.00
Vectorization ratio other	30.00
Vector-efficiency ratio all	93.92
Vector-efficiency ratio load	78.57
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	100.00
Vector-efficiency ratio add_sub	85.94
Vector-efficiency ratio fma	100.00
Vector-efficiency ratio div_sqrt	100.00
Vector-efficiency ratio other	100.00

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	1.05
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	3.05
Bottlenecks	P6, P7,
Function	.omp_outlined.#0x41c480
Source	advec_mom.cpp:45-48,context.h:69-69
Source loop unroll info	unrolled by 4
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	4
CQA cycles	14.50
CQA cycles if no scalar integer	14.50
CQA cycles if FP arith vectorized	14.50
CQA cycles if fully vectorized	13.75
Front-end cycles	4.75
DIV/SQRT cycles	0.50
P0 cycles	0.50
P1 cycles	0.25
P2 cycles	0.25
P3 cycles	0.25
P4 cycles	0.25
P5 cycles	14.50
P6 cycles	14.50
P7 cycles	2.00
P8 cycles	2.00
P9 cycles	3.50
P10 cycles	3.50
P11 cycles	0.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	1.00 - 0.50
Inter-iter dependencies cycles	2
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	38.00
Nb uops	38.00
Nb loads	NA
Nb stores	2.00
Nb stack references	0.00
FLOP/cycle	1.10
Nb FLOP add-sub	16.00
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	15.45
Bytes prefetched	0.00
Bytes loaded	160.00
Bytes stored	64.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	72.97
Vectorization ratio load	71.43
Vectorization ratio store	100.00
Vectorization ratio mul	100.00
Vectorization ratio add_sub	81.25
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	100.00
Vectorization ratio other	30.00
Vector-efficiency ratio all	93.92
Vector-efficiency ratio load	78.57
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	100.00
Vector-efficiency ratio add_sub	85.94
Vector-efficiency ratio fma	100.00
Vector-efficiency ratio div_sqrt	100.00
Vector-efficiency ratio other	100.00

Path /

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

Function	.omp_outlined.#0x41c480
Source file and lines	advec_mom.cpp:45-48
Module	exec

The loop is defined in:

/home/hbollore/qaas-runs/170-290-5445/intel/CloverLeafCXX/build/CloverLeafCXX/src/omp/advec_mom.cpp:45-48
/home/hbollore/qaas-runs/170-290-5445/intel/CloverLeafCXX/build/CloverLeafCXX/src/omp/context.h:69

The related source loop is unrolled by 4 (including vectorization).

gain
potential
hint
expert

Vectorization

Your loop is partially vectorized. 93% of vector register length is used (average across all VPU instructions). By fully vectorizing your loop, you can lower the cost of an iteration from 14.50 to 13.75 cycles (1.05x speedup).

Details

72% of VPU instructions are used in vector version (process two or more data elements in vector registers):

71% of VPU loads are used in vector version.
81% of VPU addition or subtraction instructions are used in vector version.
30% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.

Since your execution units are vector units, only a fully vectorized loop can use their full power.

Workaround

Try another compiler or update/tune your current one
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 is composed of 16 FP arithmetical operations:

16: addition or subtraction

The binary loop is loading 160 bytes. The binary loop is storing 64 bytes.

Arithmetic intensity

Arithmetic intensity is 0.07 FP operations per loaded or stored byte.

General properties

nb instructions	38
loop length	152
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

4.75 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	0.50	0.50	0.25	0.25	0.25	0.25	14.50	14.50	1.00	1.00	3.50	3.50	0.00	0.00	0.00
cycles	0.50	0.50	0.25	0.25	0.25	0.25	14.50	14.50	2.00	2.00	3.50	3.50	0.00	0.00	0.00

Cycles executing div or sqrt instructions	1.00-0.50
Longest recurrence chain latency (RecMII)	2.00

Cycles summary

Front-end	4.75
Data deps.	2.00
Overall L1	14.50

Vectorization ratios

INT

all	69%
load	71%
store	100%
mul	100%
add-sub	75%
fma	100%
other	36%

FP

all	100%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	100%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

INT+FP

all	72%
load	71%
store	100%
mul	100%
add-sub	81%
fma	100%
div/sqrt	100%
other	30%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 14.50 cycles. At this rate:

11% of peak load performance is reached (11.03 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
9% of peak store performance is reached (4.41 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 41c5e0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	Latency	Recip. throughput
MOVPRFX Z20, Z0	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
SDIV Z20.D, P0/M, Z20.D, Z2.D	1	0	0	0	0	0	0	1	0	0	0	0	0	7-20	1-0.50
ADD Z21.D, Z3.D, Z20.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MOVPRFX Z22, Z0	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MLS Z22.D, P0/M, Z20.D, Z2.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
ADD Z20.D, Z6.D, Z20.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
SUBS X12, X12, X10	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.33
ADD Z23.D, Z4.D, Z22.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MOVPRFX Z24, Z23	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
SXTW Z24.D, P0/M, Z23.D	1	0	0	0	0	0	0	0	1	0	0	0	0	2	1
ADD Z22.D, Z17.D, Z22.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
ADD Z0.D, Z0.D, Z1.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
SXTW Z20.D, P0/M, Z20.D	1	0	0	0	0	0	0	0	1	0	0	0	0	2	1
SXTW Z21.D, P0/M, Z21.D	1	0	0	0	0	0	0	0	1	0	0	0	0	2	1
MOVPRFX Z25, Z24	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MLA Z25.D, P0/M, Z5.D, Z21.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
LD1D {Z25.D}, P0/Z, [X24, Z25.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
MOVPRFX Z26, Z18	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MUL Z26.D, P0/M, Z26.D, Z21.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
ADR Z22.D, [Z26, Z22.D,SXTW]	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MAD Z20.D, P0/M, Z7.D, Z24.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
LD1D {Z20.D}, P0/Z, [X26, Z20.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FADD Z20.D, Z20.D, Z25.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MOVPRFX Z25, Z24	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MLA Z25.D, P0/M, Z7.D, Z21.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
LD1D {Z25.D}, P0/Z, [X26, Z25.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FSUB Z20.D, Z20.D, Z25.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MOVPRFX Z25, Z24	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MLA Z25.D, P0/M, Z16.D, Z21.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
MAD Z21.D, P0/M, Z19.D, Z24.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
ST1D {Z20.D}, P0, [X22, Z25.D,LSL #3]	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	2	0.50
LD1D {Z22.D}, P0/Z, [X21, Z22.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FADD Z20.D, Z22.D, Z20.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
ADR Z22.D, [Z26, Z23.D,SXTW]	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
LD1D {Z22.D}, P0/Z, [X21, Z22.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FSUB Z20.D, Z20.D, Z22.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
ST1D {Z20.D}, P0, [X28, Z21.D,LSL #3]	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	2	0.50
B.NE 41c5e0 <.omp_outlined.+0x160>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	1	0.50

Function	.omp_outlined.#0x41c480
Source file and lines	advec_mom.cpp:45-48
Module	exec

The loop is defined in:

/home/hbollore/qaas-runs/170-290-5445/intel/CloverLeafCXX/build/CloverLeafCXX/src/omp/advec_mom.cpp:45-48
/home/hbollore/qaas-runs/170-290-5445/intel/CloverLeafCXX/build/CloverLeafCXX/src/omp/context.h:69

The related source loop is unrolled by 4 (including vectorization).

gain
potential
hint
expert

Vectorization

Your loop is partially vectorized. 93% of vector register length is used (average across all VPU instructions). By fully vectorizing your loop, you can lower the cost of an iteration from 14.50 to 13.75 cycles (1.05x speedup).

Details

72% of VPU instructions are used in vector version (process two or more data elements in vector registers):

71% of VPU loads are used in vector version.
81% of VPU addition or subtraction instructions are used in vector version.
30% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.

Since your execution units are vector units, only a fully vectorized loop can use their full power.

Workaround

Try another compiler or update/tune your current one
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 is composed of 16 FP arithmetical operations:

16: addition or subtraction

The binary loop is loading 160 bytes. The binary loop is storing 64 bytes.

Arithmetic intensity

Arithmetic intensity is 0.07 FP operations per loaded or stored byte.

General properties

nb instructions	38
loop length	152
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

4.75 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	0.50	0.50	0.25	0.25	0.25	0.25	14.50	14.50	1.00	1.00	3.50	3.50	0.00	0.00	0.00
cycles	0.50	0.50	0.25	0.25	0.25	0.25	14.50	14.50	2.00	2.00	3.50	3.50	0.00	0.00	0.00

Cycles executing div or sqrt instructions	1.00-0.50
Longest recurrence chain latency (RecMII)	2.00

Cycles summary

Front-end	4.75
Data deps.	2.00
Overall L1	14.50

Vectorization ratios

INT

all	69%
load	71%
store	100%
mul	100%
add-sub	75%
fma	100%
other	36%

FP

all	100%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	100%
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

INT+FP

all	72%
load	71%
store	100%
mul	100%
add-sub	81%
fma	100%
div/sqrt	100%
other	30%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 14.50 cycles. At this rate:

11% of peak load performance is reached (11.03 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
9% of peak store performance is reached (4.41 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 41c5e0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	Latency	Recip. throughput
MOVPRFX Z20, Z0	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
SDIV Z20.D, P0/M, Z20.D, Z2.D	1	0	0	0	0	0	0	1	0	0	0	0	0	7-20	1-0.50
ADD Z21.D, Z3.D, Z20.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MOVPRFX Z22, Z0	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MLS Z22.D, P0/M, Z20.D, Z2.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
ADD Z20.D, Z6.D, Z20.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
SUBS X12, X12, X10	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.33
ADD Z23.D, Z4.D, Z22.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MOVPRFX Z24, Z23	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
SXTW Z24.D, P0/M, Z23.D	1	0	0	0	0	0	0	0	1	0	0	0	0	2	1
ADD Z22.D, Z17.D, Z22.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
ADD Z0.D, Z0.D, Z1.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
SXTW Z20.D, P0/M, Z20.D	1	0	0	0	0	0	0	0	1	0	0	0	0	2	1
SXTW Z21.D, P0/M, Z21.D	1	0	0	0	0	0	0	0	1	0	0	0	0	2	1
MOVPRFX Z25, Z24	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MLA Z25.D, P0/M, Z5.D, Z21.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
LD1D {Z25.D}, P0/Z, [X24, Z25.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
MOVPRFX Z26, Z18	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MUL Z26.D, P0/M, Z26.D, Z21.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
ADR Z22.D, [Z26, Z22.D,SXTW]	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MAD Z20.D, P0/M, Z7.D, Z24.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
LD1D {Z20.D}, P0/Z, [X26, Z20.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FADD Z20.D, Z20.D, Z25.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MOVPRFX Z25, Z24	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MLA Z25.D, P0/M, Z7.D, Z21.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
LD1D {Z25.D}, P0/Z, [X26, Z25.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FSUB Z20.D, Z20.D, Z25.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MOVPRFX Z25, Z24	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
MLA Z25.D, P0/M, Z16.D, Z21.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
MAD Z21.D, P0/M, Z19.D, Z24.D	1	0	0	0	0	0	0	1	0	0	0	0	0	5	2
ST1D {Z20.D}, P0, [X22, Z25.D,LSL #3]	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	2	0.50
LD1D {Z22.D}, P0/Z, [X21, Z22.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FADD Z20.D, Z22.D, Z20.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
ADR Z22.D, [Z26, Z23.D,SXTW]	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
LD1D {Z22.D}, P0/Z, [X21, Z22.D,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FSUB Z20.D, Z20.D, Z22.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
ST1D {Z20.D}, P0, [X28, Z21.D,LSL #3]	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	2	0.50
B.NE 41c5e0 <.omp_outlined.+0x160>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	1	0.50

Report Configuration

Vectorization

Details

Workaround

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

Arithmetic intensity

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Cycles and memory resources usage

ASM code

Vectorization

Details

Workaround

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

Arithmetic intensity

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Cycles and memory resources usage

ASM code