OV - exec - Loop 833

	Loop Id: 833	Module: exec	Source: MultiBsplineRef.hpp:242-262	Coverage: 8.86%

0x430dcc LDR	Q28, [X15, X4]    [10]

0x430dd0 LDR	Q29, [X17, X4]    [12]

0x430dd4 LDR	Q26, [X18, X4]    [13]

0x430dd8 ADD	X8, X21, X4

0x430ddc SUBS	X14, X14, #2

0x430de0 ADD	X9, X8, X1

0x430de4 LDR	Q27, [X0, X4]    [11]

0x430de8 ADD	X11, X9, X1

0x430dec LDR	Q30, [X11]    [5]

0x430df0 FMUL	V31.2D, V28.2D, V13.D[0]

0x430df4 FMUL	V9.2D, V29.2D, V3.D[0]

0x430df8 FMLA	V9.2D, V27.2D, V4.D[0]

0x430dfc FMLA	V31.2D, V26.2D, V14.D[0]

0x430e00 FADD	V31.2D, V31.2D, V9.2D

0x430e04 FMUL	V9.2D, V28.2D, V24.D[0]

0x430e08 FMUL	V28.2D, V28.2D, V15.D[0]

0x430e0c FMLA	V9.2D, V29.2D, V25.D[0]

0x430e10 FMLA	V28.2D, V29.2D, V10.D[0]

0x430e14 LDR	Q29, [X8]    [1]

0x430e18 FMLA	V30.2D, V31.2D, V17.D[0]

0x430e1c FMLA	V9.2D, V26.2D, V23.D[0]

0x430e20 FMLA	V28.2D, V26.2D, V11.D[0]

0x430e24 LDR	Q26, [X9]    [2]

0x430e28 FMLA	V28.2D, V27.2D, V12.D[0]

0x430e2c FMLA	V9.2D, V27.2D, V22.D[0]

0x430e30 STR	Q30, [X11]    [5]

0x430e34 ADD	X11, X11, X1

0x430e38 ADD	X12, X11, X1

0x430e3c LDR	Q27, [X11]    [4]

0x430e40 LDR	Q30, [X12, X1]    [6]

0x430e44 FMLA	V26.2D, V28.2D, V19.D[0]

0x430e48 FMLA	V29.2D, V28.2D, V16.D[0]

0x430e4c STR	Q26, [X9]    [2]

0x430e50 LDR	Q26, [X20, X4]    [3]

0x430e54 FMLA	V27.2D, V28.2D, V21.D[0]

0x430e58 STR	Q29, [X8]    [1]

0x430e5c FADD	V30.2D, V30.2D, V9.2D

0x430e60 LDR	Q9, [X12]    [6]

0x430e64 STR	Q30, [X12, X1]    [6]

0x430e68 STR	Q27, [X11]    [4]

0x430e6c FMLA	V26.2D, V28.2D, V17.D[0]

0x430e70 FMLA	V9.2D, V31.2D, V20.D[0]

0x430e74 STR	Q26, [X20, X4]    [3]

0x430e78 LDR	Q26, [X23, X4]    [8]

0x430e7c STR	Q9, [X12]    [6]

0x430e80 FMLA	V26.2D, V28.2D, V20.D[0]

0x430e84 STR	Q26, [X23, X4]    [8]

0x430e88 LDR	Q26, [X24, X4]    [9]

0x430e8c FMLA	V26.2D, V31.2D, V18.D[0]

0x430e90 STR	Q26, [X24, X4]    [9]

0x430e94 LDR	Q26, [X22, X4]    [7]

0x430e98 FMLA	V26.2D, V28.2D, V18.D[0]

0x430e9c STR	Q26, [X22, X4]    [7]

0x430ea0 ADD	X4, X4, #16

0x430ea4 B.NE	430dcc

/home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp: 242 - 262

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

242:       for (int n = 0; n < iSplitPoint; n++)

243:       {

244:         T coefsv    = coefs[n];

245:         T coefsvzs  = coefszs[n];

246:         T coefsv2zs = coefs2zs[n];

247:         T coefsv3zs = coefs3zs[n];

248:

249:         T sum0 = c[0] * coefsv + c[1] * coefsvzs + c[2] * coefsv2zs + c[3] * coefsv3zs;

250:         T sum1 = dc[0] * coefsv + dc[1] * coefsvzs + dc[2] * coefsv2zs + dc[3] * coefsv3zs;

251:         T sum2 = d2c[0] * coefsv + d2c[1] * coefsvzs + d2c[2] * coefsv2zs + d2c[3] * coefsv3zs;

252:

253:         hxx[n] += pre20 * sum0;

254:         hxy[n] += pre11 * sum0;

255:         hxz[n] += pre10 * sum1;

256:         hyy[n] += pre02 * sum0;

257:         hyz[n] += pre01 * sum1;

258:         hzz[n] += pre00 * sum2;

259:         gx[n] += pre10 * sum0;

260:         gy[n] += pre01 * sum0;

261:         gz[n] += pre00 * sum1;

262:         vals[n] += pre00 * sum0;

Coverage (%)	Name	Source Location	Module
►84.23+	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:206	exec
○	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:219	exec
○	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:100	exec
○	qmcplusplus::WaveFunction::rat[...]	stl_vector.h:993	exec
○	main.omp_outlined.63	miniqmc.cpp:442	exec
○	__kmp_invoke_microtask		libomp.so
►7.93+	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:206	exec
○	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:219	exec
○	qmcplusplus::SPOSet::evaluate_[...]	OhmmsVector.h:210	exec
○	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:263	exec
○	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:238	exec
○	qmcplusplus::WaveFunction::eva[...]	WaveFunction.cpp:171	exec
○	main.omp_outlined	miniqmc.cpp:379	exec
○	__kmp_invoke_microtask		libomp.so
►7.84+	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:206	exec
○	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:219	exec
○	qmcplusplus::SPOSet::evaluate_[...]	OhmmsVector.h:210	exec
○	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:263	exec
○	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:238	exec
○	qmcplusplus::WaveFunction::eva[...]	WaveFunction.cpp:171	exec
○	main.omp_outlined	miniqmc.cpp:379	exec
○	__kmp_invoke_microtask		libomp.so

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	1.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.03
Bottlenecks	P8, P9, P10, P11,
Function	void miniqmcreference::MultiBsplineEvalRef::evaluate_vgh(qmcplusplus::bspline_traits::SplineType const, double, double, double, double, double, double, unsigned long)
Source	MultiBsplineRef.hpp:242-262
Source loop unroll info	unrolled by 2
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	2
CQA cycles	8.25
CQA cycles if no scalar integer	8.25
CQA cycles if FP arith vectorized	8.25
CQA cycles if fully vectorized	8.25
Front-end cycles	6.88
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	1.25
P3 cycles	1.25
P4 cycles	1.17
P5 cycles	1.17
P6 cycles	1.08
P7 cycles	1.08
P8 cycles	8.25
P9 cycles	8.25
P10 cycles	8.25
P11 cycles	8.25
P12 cycles	8.00
P13 cycles	8.00
P14 cycles	8.00
P15 cycles	0.00
P16 cycles	0.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	55.00
Nb uops	55.00
Nb loads	NA
Nb stores	10.00
Nb stack references	0.00
FLOP/cycle	9.70
Nb FLOP add-sub	4.00
Nb FLOP mul	8.00
Nb FLOP fma	34.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	46.55
Bytes prefetched	0.00
Bytes loaded	224.00
Bytes stored	160.00
Stride 0	0.00
Stride 1	8.00
Stride n	0.00
Stride unknown	5.00
Stride indirect	0.00
Vectorization ratio all	97.92
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	100.00
Vectorization ratio add_sub	66.67
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	98.96
Vector-efficiency ratio load	100.00
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	100.00
Vector-efficiency ratio add_sub	83.33
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	1.03
Bottlenecks	P8, P9, P10, P11,
Function	void miniqmcreference::MultiBsplineEvalRef::evaluate_vgh(qmcplusplus::bspline_traits::SplineType const, double, double, double, double, double, double, unsigned long)
Source	MultiBsplineRef.hpp:242-262
Source loop unroll info	unrolled by 2
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	2
CQA cycles	8.25
CQA cycles if no scalar integer	8.25
CQA cycles if FP arith vectorized	8.25
CQA cycles if fully vectorized	8.25
Front-end cycles	6.88
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	1.25
P3 cycles	1.25
P4 cycles	1.17
P5 cycles	1.17
P6 cycles	1.08
P7 cycles	1.08
P8 cycles	8.25
P9 cycles	8.25
P10 cycles	8.25
P11 cycles	8.25
P12 cycles	8.00
P13 cycles	8.00
P14 cycles	8.00
P15 cycles	0.00
P16 cycles	0.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	55.00
Nb uops	55.00
Nb loads	NA
Nb stores	10.00
Nb stack references	0.00
FLOP/cycle	9.70
Nb FLOP add-sub	4.00
Nb FLOP mul	8.00
Nb FLOP fma	34.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	46.55
Bytes prefetched	0.00
Bytes loaded	224.00
Bytes stored	160.00
Stride 0	0.00
Stride 1	8.00
Stride n	0.00
Stride unknown	5.00
Stride indirect	0.00
Vectorization ratio all	97.92
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	100.00
Vectorization ratio add_sub	66.67
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	98.96
Vector-efficiency ratio load	100.00
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	100.00
Vector-efficiency ratio add_sub	83.33
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	void miniqmcreference::MultiBsplineEvalRef::evaluate_vgh(qmcplusplus::bspline_traits::SplineType const, double, double, double, double, double, double, unsigned long)
Source file and lines	MultiBsplineRef.hpp:242-262
Module	exec

The loop is defined in /home/hbollore/qaas/qaas-runs/174-135-6342/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp:242-262.

The related source loop is unrolled by 2 (including vectorization).

gain
potential
hint
expert

Vectorization

Your loop is highly vectorized. 98% of vector register length is used (average across all VPU instructions).

Details

97% of VPU instructions are used in vector version (process two or more data elements in vector registers):

66% of VPU addition or subtraction instructions are used in vector version.

FMA

Detected 34 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 80 FP arithmetical operations:

38: addition or subtraction (34 inside FMA instructions)
42: multiply (34 inside FMA instructions)

The binary loop is loading 224 bytes. The binary loop is storing 160 bytes.

Arithmetic intensity

Arithmetic intensity is 0.21 FP operations per loaded or stored byte.

General properties

nb instructions	55
loop length	220
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

6.88 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16
uops	0.50	0.50	1.25	1.25	1.17	1.17	1.08	1.08	8.25	8.25	8.25	8.25	8.00	8.00	8.00	0.00	0.00
cycles	0.50	0.50	1.25	1.25	1.17	1.17	1.08	1.08	8.25	8.25	8.25	8.25	8.00	8.00	8.00	0.00	0.00

Execution ports to units layout:

P0: BRU
P1: BRU
P2: ALU
P3: ALU
P4: ALU
P5: ALU
P6: ALU
P7: ALU
P8 (128 bits): VPU, FP store data, ALU, DIV/SQRT
P9 (128 bits): VPU, ALU, FP store data
P10 (128 bits): VPU, ALU, DIV/SQRT
P11 (128 bits): ALU, VPU
P12 (256 bits): store address, load
P13 (256 bits): store address, load
P14 (256 bits): load
P15 (64 bits): store data
P16 (64 bits): store data

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

Cycles summary

Front-end	6.88
Data deps.	1.00
Overall L1	8.25

Vectorization ratios

INT

all	96%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	0%
fma	NA (no fma vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

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	97%
load	100%
store	100%
mul	100%
add-sub	66%
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 8.25 cycles. At this rate:

28% of peak load performance is reached (27.15 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
40% of peak store performance is reached (19.39 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 430dcc

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	Latency	Recip. throughput	Vectorization
LDR Q28, [X15, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
LDR Q29, [X17, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
LDR Q26, [X18, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
ADD X8, X21, X4	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
SUBS X14, X14, #2	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	1	0.33	scal (50.0%)
ADD X9, X8, X1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
LDR Q27, [X0, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
ADD X11, X9, X1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
LDR Q30, [X11]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMUL V31.2D, V28.2D, V13.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	3	0.25	vect (100.0%)
FMUL V9.2D, V29.2D, V3.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	3	0.25	vect (100.0%)
FMLA V9.2D, V27.2D, V4.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V31.2D, V26.2D, V14.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FADD V31.2D, V31.2D, V9.2D	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	2	0.25	vect (100.0%)
FMUL V9.2D, V28.2D, V24.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	3	0.25	vect (100.0%)
FMUL V28.2D, V28.2D, V15.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	3	0.25	vect (100.0%)
FMLA V9.2D, V29.2D, V25.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V28.2D, V29.2D, V10.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
LDR Q29, [X8]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V30.2D, V31.2D, V17.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V9.2D, V26.2D, V23.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V28.2D, V26.2D, V11.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
LDR Q26, [X9]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V28.2D, V27.2D, V12.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V9.2D, V27.2D, V22.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q30, [X11]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
ADD X11, X11, X1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
ADD X12, X11, X1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
LDR Q27, [X11]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
LDR Q30, [X12, X1]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V26.2D, V28.2D, V19.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V29.2D, V28.2D, V16.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X9]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
LDR Q26, [X20, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V27.2D, V28.2D, V21.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q29, [X8]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
FADD V30.2D, V30.2D, V9.2D	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	2	0.25	vect (100.0%)
LDR Q9, [X12]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
STR Q30, [X12, X1]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
STR Q27, [X11]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
FMLA V26.2D, V28.2D, V17.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V9.2D, V31.2D, V20.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X20, X4]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
LDR Q26, [X23, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
STR Q9, [X12]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
FMLA V26.2D, V28.2D, V20.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X23, X4]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
LDR Q26, [X24, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V26.2D, V31.2D, V18.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X24, X4]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
LDR Q26, [X22, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V26.2D, V28.2D, V18.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X22, X4]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
ADD X4, X4, #16	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
B.NE 430dcc <_ZN16miniqmcreference19MultiBsplineEvalRef12evaluate_vghIdEEvPKN11qmcplusplus14bspline_traitsIT_Lj3EE10SplineTypeES4_S4_S4_PS4_S9_S9_m+0x76c>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A

Function	void miniqmcreference::MultiBsplineEvalRef::evaluate_vgh(qmcplusplus::bspline_traits::SplineType const, double, double, double, double, double, double, unsigned long)
Source file and lines	MultiBsplineRef.hpp:242-262
Module	exec

gain
potential
hint
expert

Vectorization

Your loop is highly vectorized. 98% of vector register length is used (average across all VPU instructions).

Details

97% of VPU instructions are used in vector version (process two or more data elements in vector registers):

66% of VPU addition or subtraction instructions are used in vector version.

FMA

Detected 34 FMA (fused multiply-add) operations. Presence of both ADD/SUB and MUL operations.

Workaround

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

The binary loop is composed of 80 FP arithmetical operations:

38: addition or subtraction (34 inside FMA instructions)
42: multiply (34 inside FMA instructions)

The binary loop is loading 224 bytes. The binary loop is storing 160 bytes.

Arithmetic intensity

Arithmetic intensity is 0.21 FP operations per loaded or stored byte.

General properties

nb instructions	55
loop length	220
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

6.88 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	P16
uops	0.50	0.50	1.25	1.25	1.17	1.17	1.08	1.08	8.25	8.25	8.25	8.25	8.00	8.00	8.00	0.00	0.00
cycles	0.50	0.50	1.25	1.25	1.17	1.17	1.08	1.08	8.25	8.25	8.25	8.25	8.00	8.00	8.00	0.00	0.00

Execution ports to units layout:

P0: BRU
P1: BRU
P2: ALU
P3: ALU
P4: ALU
P5: ALU
P6: ALU
P7: ALU
P8 (128 bits): VPU, FP store data, ALU, DIV/SQRT
P9 (128 bits): VPU, ALU, FP store data
P10 (128 bits): VPU, ALU, DIV/SQRT
P11 (128 bits): ALU, VPU
P12 (256 bits): store address, load
P13 (256 bits): store address, load
P14 (256 bits): load
P15 (64 bits): store data
P16 (64 bits): store data

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

Cycles summary

Front-end	6.88
Data deps.	1.00
Overall L1	8.25

Vectorization ratios

INT

all	96%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	0%
fma	NA (no fma vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

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	97%
load	100%
store	100%
mul	100%
add-sub	66%
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 8.25 cycles. At this rate:

28% of peak load performance is reached (27.15 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
40% of peak store performance is reached (19.39 out of 48.00 bytes stored per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 430dcc

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	Latency	Recip. throughput	Vectorization
LDR Q28, [X15, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
LDR Q29, [X17, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
LDR Q26, [X18, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
ADD X8, X21, X4	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
SUBS X14, X14, #2	1	0	0	0.25	0.25	0	0	0.25	0.25	0	0	0	0	0	0	0	1	0.33	scal (50.0%)
ADD X9, X8, X1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
LDR Q27, [X0, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
ADD X11, X9, X1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
LDR Q30, [X11]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMUL V31.2D, V28.2D, V13.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	3	0.25	vect (100.0%)
FMUL V9.2D, V29.2D, V3.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	3	0.25	vect (100.0%)
FMLA V9.2D, V27.2D, V4.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V31.2D, V26.2D, V14.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FADD V31.2D, V31.2D, V9.2D	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	2	0.25	vect (100.0%)
FMUL V9.2D, V28.2D, V24.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	3	0.25	vect (100.0%)
FMUL V28.2D, V28.2D, V15.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	3	0.25	vect (100.0%)
FMLA V9.2D, V29.2D, V25.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V28.2D, V29.2D, V10.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
LDR Q29, [X8]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V30.2D, V31.2D, V17.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V9.2D, V26.2D, V23.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V28.2D, V26.2D, V11.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
LDR Q26, [X9]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V28.2D, V27.2D, V12.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V9.2D, V27.2D, V22.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q30, [X11]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
ADD X11, X11, X1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
ADD X12, X11, X1	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
LDR Q27, [X11]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
LDR Q30, [X12, X1]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V26.2D, V28.2D, V19.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V29.2D, V28.2D, V16.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X9]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
LDR Q26, [X20, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V27.2D, V28.2D, V21.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q29, [X8]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
FADD V30.2D, V30.2D, V9.2D	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	2	0.25	vect (100.0%)
LDR Q9, [X12]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
STR Q30, [X12, X1]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
STR Q27, [X11]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
FMLA V26.2D, V28.2D, V17.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
FMLA V9.2D, V31.2D, V20.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X20, X4]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
LDR Q26, [X23, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
STR Q9, [X12]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
FMLA V26.2D, V28.2D, V20.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X23, X4]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
LDR Q26, [X24, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V26.2D, V31.2D, V18.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X24, X4]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
LDR Q26, [X22, X4]	1	0	0	0	0	0	0	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33	vect (100.0%)
FMLA V26.2D, V28.2D, V18.D[0]	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	4	0.25	vect (100.0%)
STR Q26, [X22, X4]	1	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0.50	0.50	0	2	0.50	vect (100.0%)
ADD X4, X4, #16	1	0	0	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	1	0.17	N/A
B.NE 430dcc <_ZN16miniqmcreference19MultiBsplineEvalRef12evaluate_vghIdEEvPKN11qmcplusplus14bspline_traitsIT_Lj3EE10SplineTypeES4_S4_S4_PS4_S9_S9_m+0x76c>	1	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0.50	N/A

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