OV - exec - Loop 755

0x43e8b0 LD1D	{Z7.D}, P0/Z, [X17, X30,LSL #3]    [1]

0x43e8b4 LD1D	{Z16.D}, P0/Z, [X16, X30,LSL #3]    [3]

0x43e8b8 LD1D	{Z2.D}, P0/Z, [X21, X30,LSL #3]    [5]

0x43e8bc LD1D	{Z19.D}, P0/Z, [X18, X30,LSL #3]    [4]

0x43e8c0 LD1D	{Z20.D}, P0/Z, [X3, X30,LSL #3]    [2]

0x43e8c4 FMUL	Z1.D, Z23.D, Z7.D

0x43e8c8 FMUL	Z17.D, Z21.D, Z2.D

0x43e8cc FMAD	Z16.D, P1/M, Z5.D, Z1.D

0x43e8d0 FMAD	Z19.D, P1/M, Z18.D, Z17.D

0x43e8d4 FADD	Z22.D, Z16.D, Z19.D

0x43e8d8 FMAD	Z22.D, P1/M, Z29.D, Z20.D

0x43e8dc ST1D	{Z22.D}, P0, [X3, X30,LSL #3]    [2]

0x43e8e0 ADD	X30, X30, X24

0x43e8e4 WHILELO	P0.D, X30, X19

0x43e8e8 B.NE	43e8b0

/home/hbollore/qaas/qaas-runs/169-816-7419/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp: 72 - 74

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

72:         vals[n] += pre00 *

73:             (c[0] * coefs[n] + c[1] * coefs[n + zs] + c[2] * coefs[n + 2 * zs] +

74:              c[3] * coefs[n + 3 * zs]);

Coverage (%)	Name	Source Location	Module
►26.09+	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:194	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:214	exec
○	main._omp_fn.1	refwrap.h:337	exec
○	GOMP_parallel		libomp.so
○	main	miniqmc.cpp:404	exec
○	__libc_start_main		libc-2.31.so
○	_start	miniqmc.cpp:546	exec
►21.74+	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:194	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:214	exec
○	main._omp_fn.1	refwrap.h:337	exec
○	GOMP_parallel		libomp.so
○	main	miniqmc.cpp:404	exec
○	__libc_start_main		libc-2.31.so
○	_start	miniqmc.cpp:546	exec
►17.39+	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:194	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:214	exec
○	main._omp_fn.1	refwrap.h:337	exec
○	GOMP_parallel		libomp.so
○	main	miniqmc.cpp:404	exec
○	__libc_start_main		libc-2.31.so
○	_start	miniqmc.cpp:546	exec
►13.04+	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:194	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:214	exec
○	main._omp_fn.1	refwrap.h:337	exec
○	GOMP_parallel		libomp.so
○	main	miniqmc.cpp:404	exec
○	__libc_start_main		libc-2.31.so
○	_start	miniqmc.cpp:546	exec
►8.70+	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:194	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:214	exec
○	main._omp_fn.1	refwrap.h:337	exec
○	GOMP_parallel		libomp.so
○	main	miniqmc.cpp:402	exec
○	__libc_start_main		libc-2.31.so
○	_start	miniqmc.cpp:546	exec
►4.35+	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:194	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:214	exec
○	main._omp_fn.1	refwrap.h:337	exec
○	GOMP_parallel		libomp.so
○	main	miniqmc.cpp:402	exec
○	__libc_start_main		libc-2.31.so
○	_start	miniqmc.cpp:546	exec
►4.35+	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:194	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:214	exec
○	main._omp_fn.1	refwrap.h:337	exec
○	GOMP_parallel		libomp.so
○	main	miniqmc.cpp:402	exec
○	__libc_start_main		libc-2.31.so
○	_start	miniqmc.cpp:546	exec
►4.35+	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:194	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:214	exec
○	main._omp_fn.1	refwrap.h:337	exec
○	GOMP_parallel		libomp.so
○	main	miniqmc.cpp:402	exec
○	__libc_start_main		libc-2.31.so
○	_start	miniqmc.cpp:546	exec

Path /

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	1.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.60
Bottlenecks	P6, P7, P8, P9, P10, P11,
Function	miniqmcreference::einspline_spo_ref::evaluate(qmcplusplus::ParticleSet const&, int, qmcplusplus::Vector >&)
Source	MultiBsplineRef.hpp:72-74
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	3.00
CQA cycles if no scalar integer	3.00
CQA cycles if FP arith vectorized	3.00
CQA cycles if fully vectorized	3.00
Front-end cycles	1.88
DIV/SQRT cycles	0.50
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	0.25
P3 cycles	1.00
P4 cycles	0.25
P5 cycles	3.00
P6 cycles	3.00
P7 cycles	3.00
P8 cycles	3.00
P9 cycles	3.00
P10 cycles	3.00
P11 cycles	0.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	15.00
Nb uops	15.00
Nb loads	NA
Nb stores	1.00
Nb stack references	0.00
FLOP/cycle	12.00
Nb FLOP add-sub	4.00
Nb FLOP mul	8.00
Nb FLOP fma	12.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	64.00
Bytes prefetched	0.00
Bytes loaded	160.00
Bytes stored	32.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	100.00
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	100.00
Vectorization ratio add_sub	100.00
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	100.00
Vector-efficiency ratio load	100.00
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	100.00
Vector-efficiency ratio add_sub	100.00
Vector-efficiency ratio fma	100.00
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	1.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.60
Bottlenecks	P6, P7, P8, P9, P10, P11,
Function	miniqmcreference::einspline_spo_ref::evaluate(qmcplusplus::ParticleSet const&, int, qmcplusplus::Vector >&)
Source	MultiBsplineRef.hpp:72-74
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	3.00
CQA cycles if no scalar integer	3.00
CQA cycles if FP arith vectorized	3.00
CQA cycles if fully vectorized	3.00
Front-end cycles	1.88
DIV/SQRT cycles	0.50
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	0.25
P3 cycles	1.00
P4 cycles	0.25
P5 cycles	3.00
P6 cycles	3.00
P7 cycles	3.00
P8 cycles	3.00
P9 cycles	3.00
P10 cycles	3.00
P11 cycles	0.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	15.00
Nb uops	15.00
Nb loads	NA
Nb stores	1.00
Nb stack references	0.00
FLOP/cycle	12.00
Nb FLOP add-sub	4.00
Nb FLOP mul	8.00
Nb FLOP fma	12.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	64.00
Bytes prefetched	0.00
Bytes loaded	160.00
Bytes stored	32.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	100.00
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	100.00
Vectorization ratio add_sub	100.00
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	100.00
Vector-efficiency ratio load	100.00
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	100.00
Vector-efficiency ratio add_sub	100.00
Vector-efficiency ratio fma	100.00
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	miniqmcreference::einspline_spo_ref::evaluate(qmcplusplus::ParticleSet const&, int, qmcplusplus::Vector >&)
Source file and lines	MultiBsplineRef.hpp:72-74
Module	exec

The loop is defined in /home/hbollore/qaas/qaas-runs/169-816-7419/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp:72-74.

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

gain
potential
hint
expert

Vectorization

Your loop is fully vectorized, using full register length.

Details

All VPU instructions are used in vector version (process two or more data elements in vector registers).

FMA

Detected 12 FMA (fused multiply-add) operations. Presence of both ADD/SUB and MUL operations.

Workaround

Try to change order in which elements are evaluated (using parentheses) in arithmetic expressions containing both ADD/SUB and MUL operations to enable your compiler to generate FMA instructions wherever possible. For instance a + b*c is a valid FMA (MUL then ADD). However (a+b)* c cannot be translated into an FMA (ADD then MUL).

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

The binary loop is composed of 36 FP arithmetical operations:

16: addition or subtraction (12 inside FMA instructions)
20: multiply (12 inside FMA instructions)

The binary loop is loading 160 bytes. The binary loop is storing 32 bytes.

Arithmetic intensity

Arithmetic intensity is 0.19 FP operations per loaded or stored byte.

General properties

nb instructions	15
loop length	60
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

1.88 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	0.50	0.50	0.50	0.25	1.00	0.25	3.00	3.00	0.50	0.50	3.00	3.00	0.00	0.00	0.00
cycles	0.50	0.50	0.50	0.25	1.00	0.25	3.00	3.00	3.00	3.00	3.00	3.00	0.00	0.00	0.00

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

Cycles summary

Front-end	1.88
Data deps.	1.00
Overall L1	3.00

Vectorization ratios

INT

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

FP

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

INT+FP

all	100%
load	100%
store	100%
mul	100%
add-sub	100%
fma	100%
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 3.00 cycles. At this rate:

55% of peak load performance is reached (53.33 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
22% of peak store performance is reached (10.67 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 43e8b0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	Latency	Recip. throughput
LD1D {Z7.D}, P0/Z, [X17, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
LD1D {Z16.D}, P0/Z, [X16, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
LD1D {Z2.D}, P0/Z, [X21, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
LD1D {Z19.D}, P0/Z, [X18, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
LD1D {Z20.D}, P0/Z, [X3, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FMUL Z1.D, Z23.D, Z7.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	3	0.50
FMUL Z17.D, Z21.D, Z2.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	3	0.50
FMAD Z16.D, P1/M, Z5.D, Z1.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50
FMAD Z19.D, P1/M, Z18.D, Z17.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50
FADD Z22.D, Z16.D, Z19.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
FMAD Z22.D, P1/M, Z29.D, Z20.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50
ST1D {Z22.D}, P0, [X3, X30,LSL #3]	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	2	0.50
ADD X30, X30, X24	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25
WHILELO P0.D, X30, X19	1	0	0	0	0	1	0	0	0	0	0	0	0	3	2
B.NE 43e8b0 <_ZN16miniqmcreference17einspline_spo_refIdE8evaluateERKN11qmcplusplus11ParticleSetEiRNS2_6VectorIdSaIdEEE+0x2f0>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	1	0.50

Function	miniqmcreference::einspline_spo_ref::evaluate(qmcplusplus::ParticleSet const&, int, qmcplusplus::Vector >&)
Source file and lines	MultiBsplineRef.hpp:72-74
Module	exec

The loop is defined in /home/hbollore/qaas/qaas-runs/169-816-7419/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp:72-74.

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

gain
potential
hint
expert

Vectorization

Your loop is fully vectorized, using full register length.

Details

All VPU instructions are used in vector version (process two or more data elements in vector registers).

FMA

Detected 12 FMA (fused multiply-add) operations. Presence of both ADD/SUB and MUL operations.

Workaround

Try to change order in which elements are evaluated (using parentheses) in arithmetic expressions containing both ADD/SUB and MUL operations to enable your compiler to generate FMA instructions wherever possible. For instance a + b*c is a valid FMA (MUL then ADD). However (a+b)* c cannot be translated into an FMA (ADD then MUL).

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

The binary loop is composed of 36 FP arithmetical operations:

16: addition or subtraction (12 inside FMA instructions)
20: multiply (12 inside FMA instructions)

The binary loop is loading 160 bytes. The binary loop is storing 32 bytes.

Arithmetic intensity

Arithmetic intensity is 0.19 FP operations per loaded or stored byte.

General properties

nb instructions	15
loop length	60
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

1.88 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	0.50	0.50	0.50	0.25	1.00	0.25	3.00	3.00	0.50	0.50	3.00	3.00	0.00	0.00	0.00
cycles	0.50	0.50	0.50	0.25	1.00	0.25	3.00	3.00	3.00	3.00	3.00	3.00	0.00	0.00	0.00

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

Cycles summary

Front-end	1.88
Data deps.	1.00
Overall L1	3.00

Vectorization ratios

INT

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

FP

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

INT+FP

all	100%
load	100%
store	100%
mul	100%
add-sub	100%
fma	100%
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 3.00 cycles. At this rate:

55% of peak load performance is reached (53.33 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
22% of peak store performance is reached (10.67 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 43e8b0

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	Latency	Recip. throughput
LD1D {Z7.D}, P0/Z, [X17, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
LD1D {Z16.D}, P0/Z, [X16, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
LD1D {Z2.D}, P0/Z, [X21, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
LD1D {Z19.D}, P0/Z, [X18, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
LD1D {Z20.D}, P0/Z, [X3, X30,LSL #3]	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	6	0.50
FMUL Z1.D, Z23.D, Z7.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	3	0.50
FMUL Z17.D, Z21.D, Z2.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	3	0.50
FMAD Z16.D, P1/M, Z5.D, Z1.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50
FMAD Z19.D, P1/M, Z18.D, Z17.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50
FADD Z22.D, Z16.D, Z19.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	2	0.50
FMAD Z22.D, P1/M, Z29.D, Z20.D	1	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50
ST1D {Z22.D}, P0, [X3, X30,LSL #3]	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	2	0.50
ADD X30, X30, X24	1	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	1	0.25
WHILELO P0.D, X30, X19	1	0	0	0	0	1	0	0	0	0	0	0	0	3	2
B.NE 43e8b0 <_ZN16miniqmcreference17einspline_spo_refIdE8evaluateERKN11qmcplusplus11ParticleSetEiRNS2_6VectorIdSaIdEEE+0x2f0>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	1	0.50

Report Configuration

Vectorization

Details

FMA

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

FMA

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