OV - exec - Loop 999

	Loop Id: 999	Module: exec	Source: MultiBsplineRef.hpp:276-286	Coverage: 0.12%

0x487834 LEA	0x8(%RDI),%R10

0x487838 VMULSD	(%R11,%RDI,1),%XMM10,%XMM2    [2]

0x48783e VMULSD	(%R12,%RDI,1),%XMM8,%XMM5    [3]

0x487844 VMOVSD	%XMM2,(%R11,%RDI,1)    [2]

0x48784a VMOVSD	%XMM5,(%R12,%RDI,1)    [3]

0x487850 VMULSD	(%RCX,%RDI,1),%XMM11,%XMM3    [9]

0x487855 VMULSD	(%R8,%RDI,1),%XMM7,%XMM0    [1]

0x48785b VMOVSD	%XMM3,(%RCX,%RDI,1)    [9]

0x487860 VMOVSD	%XMM0,(%R8,%RDI,1)    [1]

0x487866 VMULSD	(%RBX,%RDI,1),%XMM12,%XMM9    [8]

0x48786b VMULSD	(%RAX,%RDI,1),%XMM6,%XMM1    [7]

0x487870 VMOVSD	%XMM9,(%RBX,%RDI,1)    [8]

0x487875 VMOVSD	%XMM1,(%RAX,%RDI,1)    [7]

0x48787a VMULSD	(%R15,%RDI,1),%XMM13,%XMM4    [6]

0x487880 VMULSD	0x8(%R12,%RDI,1),%XMM8,%XMM1    [3]

0x487887 VMOVSD	%XMM4,(%R15,%RDI,1)    [6]

0x48788d VMULSD	(%R14,%RDI,1),%XMM14,%XMM5    [5]

0x487893 VMOVSD	%XMM5,(%R14,%RDI,1)    [5]

0x487899 VMULSD	(%R13,%RDI,1),%XMM15,%XMM0    [4]

0x4878a0 VMOVSD	%XMM0,(%R13,%RDI,1)    [4]

0x4878a7 ADD	$0x10,%RDI

0x4878ab VMOVSD	%XMM1,(%R12,%R10,1)    [3]

0x4878b1 VMULSD	(%R11,%R10,1),%XMM10,%XMM9    [2]

0x4878b7 VMULSD	(%R8,%R10,1),%XMM7,%XMM2    [1]

0x4878bd VMOVSD	%XMM9,(%R11,%R10,1)    [2]

0x4878c3 VMOVSD	%XMM2,(%R8,%R10,1)    [1]

0x4878c9 VMULSD	(%RCX,%R10,1),%XMM11,%XMM4    [9]

0x4878cf VMULSD	(%RAX,%R10,1),%XMM6,%XMM3    [7]

0x4878d5 VMOVSD	%XMM4,(%RCX,%R10,1)    [9]

0x4878db VMOVSD	%XMM3,(%RAX,%R10,1)    [7]

0x4878e1 VMULSD	(%RBX,%R10,1),%XMM12,%XMM5    [8]

0x4878e7 VMOVSD	%XMM5,(%RBX,%R10,1)    [8]

0x4878ed VMULSD	(%R15,%R10,1),%XMM13,%XMM0    [6]

0x4878f3 VMOVSD	%XMM0,(%R15,%R10,1)    [6]

0x4878f9 VMULSD	(%R14,%R10,1),%XMM14,%XMM1    [5]

0x4878ff VMOVSD	%XMM1,(%R14,%R10,1)    [5]

0x487905 VMULSD	(%R13,%R10,1),%XMM15,%XMM2    [4]

0x48790c VMOVSD	%XMM2,(%R13,%R10,1)    [4]

0x487913 CMP	%RDI,%R9

0x487916 JNE	487834

/home/eoseret/qaas_runs_ZEN5/173-938-0042/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp: 276 - 286

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

276:   for (int n = 0; n < num_splines; n++)

277:   {

278:     gx[n] *= dxInv;

279:     gy[n] *= dyInv;

280:     gz[n] *= dzInv;

281:     hxx[n] *= dxx;

282:     hyy[n] *= dyy;

283:     hzz[n] *= dzz;

284:     hxy[n] *= dxy;

285:     hxz[n] *= dxz;

286:     hyz[n] *= dyz;

Coverage (%)	Name	Source Location	Module
►40.48+	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:206	exec
○	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:100	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:196	exec
○	main._omp_fn.1	stl_vector.h:1131	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►39.04+	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:206	exec
○	miniqmcreference::DiracDetermi[...]	DiracDeterminantRef.cpp:100	exec
○	qmcplusplus::WaveFunction::rat[...]	WaveFunction.cpp:196	exec
○	main._omp_fn.1	stl_vector.h:1131	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►8.53+	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:206	exec
○	miniqmcreference::DiracDetermi[...]	OhmmsVector.h:210	exec
○	qmcplusplus::WaveFunction::eva[...]	WaveFunction.cpp:172	exec
○	main._omp_fn.0	miniqmc.cpp:397	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►8.49+	miniqmcreference::einspline_sp[...]	einspline_spo_ref.hpp:206	exec
○	miniqmcreference::DiracDetermi[...]	OhmmsVector.h:210	exec
○	qmcplusplus::WaveFunction::eva[...]	WaveFunction.cpp:171	exec
○	main._omp_fn.0	miniqmc.cpp:397	exec
○	gomp_thread_start	team.c:130	libgomp.so.1.0.0
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6

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	8.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.85
Bottlenecks	P6, P7, P8, P9, P10, P11, P14, P15,
Function	void miniqmcreference::MultiBsplineEvalRef::evaluate_vgh(qmcplusplus::bspline_traits::SplineType const, double, double, double, double, double, double, unsigned long)
Source	MultiBsplineRef.hpp:276-286
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	9.00
CQA cycles if no scalar integer	9.00
CQA cycles if FP arith vectorized	9.00
CQA cycles if fully vectorized	1.13
Front-end cycles	4.88
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	0.50
P3 cycles	0.50
P4 cycles	0.50
P5 cycles	0.50
P6 cycles	9.00
P7 cycles	9.00
P8 cycles	9.00
P9 cycles	9.00
P10 cycles	9.00
P11 cycles	9.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	9.00
P15 cycles	9.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	40.00
Nb uops	39.00
Nb loads	18.00
Nb stores	18.00
Nb stack references	0.00
FLOP/cycle	2.00
Nb FLOP add-sub	0.00
Nb FLOP mul	18.00
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	32.00
Bytes prefetched	0.00
Bytes loaded	144.00
Bytes stored	144.00
Stride 0	0.00
Stride 1	1.00
Stride n	8.00
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	0.00
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	12.50
Vector-efficiency ratio load	12.50
Vector-efficiency ratio store	12.50
Vector-efficiency ratio mul	12.50
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	8.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.85
Bottlenecks	P6, P7, P8, P9, P10, P11, P14, P15,
Function	void miniqmcreference::MultiBsplineEvalRef::evaluate_vgh(qmcplusplus::bspline_traits::SplineType const, double, double, double, double, double, double, unsigned long)
Source	MultiBsplineRef.hpp:276-286
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	9.00
CQA cycles if no scalar integer	9.00
CQA cycles if FP arith vectorized	9.00
CQA cycles if fully vectorized	1.13
Front-end cycles	4.88
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	0.50
P3 cycles	0.50
P4 cycles	0.50
P5 cycles	0.50
P6 cycles	9.00
P7 cycles	9.00
P8 cycles	9.00
P9 cycles	9.00
P10 cycles	9.00
P11 cycles	9.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	9.00
P15 cycles	9.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	40.00
Nb uops	39.00
Nb loads	18.00
Nb stores	18.00
Nb stack references	0.00
FLOP/cycle	2.00
Nb FLOP add-sub	0.00
Nb FLOP mul	18.00
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	32.00
Bytes prefetched	0.00
Bytes loaded	144.00
Bytes stored	144.00
Stride 0	0.00
Stride 1	1.00
Stride n	8.00
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	0.00
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	NA
Vector-efficiency ratio all	12.50
Vector-efficiency ratio load	12.50
Vector-efficiency ratio store	12.50
Vector-efficiency ratio mul	12.50
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	void miniqmcreference::MultiBsplineEvalRef::evaluate_vgh(qmcplusplus::bspline_traits::SplineType const, double, double, double, double, double, double, unsigned long)
Source file and lines	MultiBsplineRef.hpp:276-286
Module	exec

The loop is defined in /home/eoseret/qaas_runs_ZEN5/173-938-0042/intel/miniqmc/build/miniqmc/src/Numerics/Spline2/MultiBsplineRef.hpp:276-286.

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

gain
potential
hint
expert

Vectorization

Your loop is not vectorized. 8 data elements could be processed at once in vector registers.

By vectorizing your loop, you can lower the cost of an iteration from 9.00 to 1.12 cycles (8.00x speedup).

Details

All SSE/AVX 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
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)

Execution units bottlenecks

Performance is limited by:

execution of FP multiply or FMA (fused multiply-add) operations (the FP multiply/FMA unit is a bottleneck)
reading data from caches/RAM (load units are a bottleneck)
writing data to caches/RAM (the store unit is a bottleneck)

Workaround

Reduce the number of FP multiply/FMA instructions
Read less array elements All SSE and/or AVX registers are used: in that case, try to relax register pressure by reducing the unroll factor or splitting your loop
Write less array elements
Provide more information to your compiler:
- hardcode the bounds of the corresponding 'for' loop

No data for this section

Slow data structures access

Detected data structures (typically arrays) that cannot be efficiently read/written

Details

Constant non-unit stride: 8 occurrence(s)

Non-unit stride (uncontiguous) accesses are not efficiently using data caches

Workaround

Try to reorganize arrays of structures to structures of arrays
Consider to permute loops (see vectorization gain report)

Type of elements and instruction set

18 SSE or AVX instructions are processing arithmetic or math operations on double precision FP elements in scalar mode (one at a time).

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

The binary loop is composed of 18 FP arithmetical operations:

18: multiply

The binary loop is loading 144 bytes (18 double precision FP elements). The binary loop is storing 144 bytes (18 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.06 FP operations per loaded or stored byte.

Unroll opportunity

Loop is data access bound.

Workaround

Unroll your loop if trip count is significantly higher than target unroll factor and if some data references are common to consecutive iterations. This can be done manually. Or by recompiling with -funroll-loops and/or -floop-unroll-and-jam. Or with the unroll (resp. unroll_and_jam) directive on top of the inner (resp. surrounding) loop. You can enforce an unroll factor: #pragma GCC unroll N

General properties

nb instructions	40
nb uops	39
loop length	232
used x86 registers	12
used mmx registers	0
used xmm registers	16
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	4.88 cycles
front end	4.88 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.50	0.50	0.50	0.50	0.50	0.50	9.00	9.00	9.00	9.00	9.00	9.00	0.00	0.00	9.00	9.00
cycles	0.50	0.50	0.50	0.50	0.50	0.50	9.00	9.00	9.00	9.00	9.00	9.00	0.00	0.00	9.00	9.00

Execution ports to units layout:

P0: ALU
P1: ALU
P2: ALU
P3: ALU, BRU
P4: ALU, BRU
P5: ALU, BRU
P6 (512 bits): store address, load
P7 (512 bits): store address, load
P8 (512 bits): store address, load
P9 (512 bits): store address, load
P10 (512 bits): VPU, DIV/SQRT
P11 (512 bits): VPU, DIV/SQRT
P12 (512 bits): VPU
P13 (512 bits): VPU
P14 (512 bits): FP store data
P15 (512 bits): FP store data

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

Cycles summary

Front-end	4.88
Dispatch	9.00
Data deps.	1.00
Overall L1	9.00

Vectorization ratios

all	0%
load	0%
store	0%
mul	0%
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	12%
load	12%
store	12%
mul	12%
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 9.00 cycles. At this rate:

6% of peak load performance is reached (16.00 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
12% of peak store performance is reached (16.00 out of 128.00 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: 487834

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P14	P15	Latency	Recip. throughput	Vectorization
LEA 0x8(%RDI),%R10	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.25	N/A
VMULSD (%R11,%RDI,1),%XMM10,%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%R12,%RDI,1),%XMM8,%XMM5	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM2,(%R11,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM5,(%R12,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%RCX,%RDI,1),%XMM11,%XMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%R8,%RDI,1),%XMM7,%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM3,(%RCX,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM0,(%R8,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%RBX,%RDI,1),%XMM12,%XMM9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%RAX,%RDI,1),%XMM6,%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM9,(%RBX,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM1,(%RAX,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R15,%RDI,1),%XMM13,%XMM4	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD 0x8(%R12,%RDI,1),%XMM8,%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM4,(%R15,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R14,%RDI,1),%XMM14,%XMM5	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM5,(%R14,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R13,%RDI,1),%XMM15,%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM0,(%R13,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
ADD $0x10,%RDI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.17	N/A
VMOVSD %XMM1,(%R12,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R11,%R10,1),%XMM10,%XMM9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%R8,%R10,1),%XMM7,%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM9,(%R11,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM2,(%R8,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%RCX,%R10,1),%XMM11,%XMM4	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%RAX,%R10,1),%XMM6,%XMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM4,(%RCX,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM3,(%RAX,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%RBX,%R10,1),%XMM12,%XMM5	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM5,(%RBX,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R15,%R10,1),%XMM13,%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM0,(%R15,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R14,%R10,1),%XMM14,%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM1,(%R14,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R13,%R10,1),%XMM15,%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM2,(%R13,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
CMP %RDI,%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.17	N/A
JNE 487834 <_ZN16miniqmcreference19MultiBsplineEvalRef12evaluate_vghIdEEvPKN11qmcplusplus14bspline_traitsIT_Lj3EE10SplineTypeES4_S4_S4_PS4_S9_S9_m+0xa24>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	1	0.50-1	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:276-286
Module	exec

gain
potential
hint
expert

Vectorization

Your loop is not vectorized. 8 data elements could be processed at once in vector registers.

By vectorizing your loop, you can lower the cost of an iteration from 9.00 to 1.12 cycles (8.00x speedup).

Details

All SSE/AVX 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
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)

Execution units bottlenecks

Performance is limited by:

execution of FP multiply or FMA (fused multiply-add) operations (the FP multiply/FMA unit is a bottleneck)
reading data from caches/RAM (load units are a bottleneck)
writing data to caches/RAM (the store unit is a bottleneck)

Workaround

Reduce the number of FP multiply/FMA instructions
Read less array elements All SSE and/or AVX registers are used: in that case, try to relax register pressure by reducing the unroll factor or splitting your loop
Write less array elements
Provide more information to your compiler:
- hardcode the bounds of the corresponding 'for' loop

No data for this section

Slow data structures access

Detected data structures (typically arrays) that cannot be efficiently read/written

Details

Constant non-unit stride: 8 occurrence(s)

Non-unit stride (uncontiguous) accesses are not efficiently using data caches

Workaround

Try to reorganize arrays of structures to structures of arrays
Consider to permute loops (see vectorization gain report)

Type of elements and instruction set

18 SSE or AVX instructions are processing arithmetic or math operations on double precision FP elements in scalar mode (one at a time).

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

The binary loop is composed of 18 FP arithmetical operations:

18: multiply

The binary loop is loading 144 bytes (18 double precision FP elements). The binary loop is storing 144 bytes (18 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.06 FP operations per loaded or stored byte.

Unroll opportunity

Loop is data access bound.

Workaround

General properties

nb instructions	40
nb uops	39
loop length	232
used x86 registers	12
used mmx registers	0
used xmm registers	16
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	4.88 cycles
front end	4.88 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.50	0.50	0.50	0.50	0.50	0.50	9.00	9.00	9.00	9.00	9.00	9.00	0.00	0.00	9.00	9.00
cycles	0.50	0.50	0.50	0.50	0.50	0.50	9.00	9.00	9.00	9.00	9.00	9.00	0.00	0.00	9.00	9.00

Execution ports to units layout:

P0: ALU
P1: ALU
P2: ALU
P3: ALU, BRU
P4: ALU, BRU
P5: ALU, BRU
P6 (512 bits): store address, load
P7 (512 bits): store address, load
P8 (512 bits): store address, load
P9 (512 bits): store address, load
P10 (512 bits): VPU, DIV/SQRT
P11 (512 bits): VPU, DIV/SQRT
P12 (512 bits): VPU
P13 (512 bits): VPU
P14 (512 bits): FP store data
P15 (512 bits): FP store data

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

Cycles summary

Front-end	4.88
Dispatch	9.00
Data deps.	1.00
Overall L1	9.00

Vectorization ratios

all	0%
load	0%
store	0%
mul	0%
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	12%
load	12%
store	12%
mul	12%
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 9.00 cycles. At this rate:

6% of peak load performance is reached (16.00 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
12% of peak store performance is reached (16.00 out of 128.00 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: 487834

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P14	P15	Latency	Recip. throughput	Vectorization
LEA 0x8(%RDI),%R10	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.25	N/A
VMULSD (%R11,%RDI,1),%XMM10,%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%R12,%RDI,1),%XMM8,%XMM5	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM2,(%R11,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM5,(%R12,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%RCX,%RDI,1),%XMM11,%XMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%R8,%RDI,1),%XMM7,%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM3,(%RCX,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM0,(%R8,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%RBX,%RDI,1),%XMM12,%XMM9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%RAX,%RDI,1),%XMM6,%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM9,(%RBX,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM1,(%RAX,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R15,%RDI,1),%XMM13,%XMM4	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD 0x8(%R12,%RDI,1),%XMM8,%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM4,(%R15,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R14,%RDI,1),%XMM14,%XMM5	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM5,(%R14,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R13,%RDI,1),%XMM15,%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM0,(%R13,%RDI,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
ADD $0x10,%RDI	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.17	N/A
VMOVSD %XMM1,(%R12,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R11,%R10,1),%XMM10,%XMM9	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%R8,%R10,1),%XMM7,%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM9,(%R11,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM2,(%R8,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%RCX,%R10,1),%XMM11,%XMM4	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMULSD (%RAX,%R10,1),%XMM6,%XMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM4,(%RCX,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMOVSD %XMM3,(%RAX,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%RBX,%R10,1),%XMM12,%XMM5	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM5,(%RBX,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R15,%R10,1),%XMM13,%XMM0	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM0,(%R15,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R14,%R10,1),%XMM14,%XMM1	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM1,(%R14,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
VMULSD (%R13,%R10,1),%XMM15,%XMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	0	0	3	0.50	scal (12.5%)
VMOVSD %XMM2,(%R13,%R10,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0.50	0.50	1	0.50	scal (12.5%)
CMP %RDI,%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	1	0.17	N/A
JNE 487834 <_ZN16miniqmcreference19MultiBsplineEvalRef12evaluate_vghIdEEvPKN11qmcplusplus14bspline_traitsIT_Lj3EE10SplineTypeES4_S4_S4_PS4_S9_S9_m+0xa24>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	1	0.50-1	N/A

Report Configuration

Vectorization

Details

Workaround

Execution units bottlenecks

Workaround

Slow data structures access

Details

Workaround

Type of elements and instruction set

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

Arithmetic intensity

Unroll opportunity

Workaround

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

Execution units bottlenecks

Workaround

Slow data structures access

Details

Workaround

Type of elements and instruction set

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

Arithmetic intensity

Unroll opportunity

Workaround

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code