OV - exec - Loop 3889

	Loop Id: 3889	Module: exec	Source: VectorSoAContainer.h:271-273 [...]	Coverage: 0.01%

0x50fd0c ORR	X0, XZR, X21

0x50fd10 BL	4c4660

0x50fd14 LDR	X25, [SP, #64]    [3]

0x50fd18 LDP	X19, X20, [SP, #16]    [3]

0x50fd1c LDP	X21, X22, [SP, #32]    [3]

0x50fd20 LDP	X23, X24, [SP, #48]    [3]

0x50fd24 LDP	X29, X30, [SP], #80    [3]

0x50fd28 RET

0x50fd2c LDR	X0, [X19, #8]    [11]

0x50fd30 MOVZ	W23, #24

0x50fd34 LDR	X24, [X22, #64]    [5]

0x50fd38 SMADDL	X22, W20, W23, XZR

0x50fd3c LDR	X0, [X0, #632]    [2]

0x50fd40 ADD	X23, X24, X22

0x50fd44 LDR	X1, [X0]    [6]

0x50fd48 LDR	X2, [X1, #72]    [9]

0x50fd4c CMP	X2, X25

0x50fd50 B.NE	50feec

0x50fd54 ADD	X0, X0, #64

0x50fd58 LDR	X14, [X19, #96]    [11]

0x50fd5c MOVZ	W25, #40

0x50fd60 SMADDL	X20, W20, W25, XZR

0x50fd64 LDR	X12, [X19, #72]    [11]

0x50fd68 LDR	X9, [X0, #8]    [7]

0x50fd6c ADD	X6, X14, X20

0x50fd70 LDR	X5, [X6, #8]    [4]

0x50fd74 ADD	X15, X12, X20

0x50fd78 LDR	X7, [X0, #24]    [7]

0x50fd7c LDR	X16, [X6, #24]    [4]

0x50fd80 LDR	D31, [X24, X22]    [8]

0x50fd84 ADD	X8, X7, X9,LSL #3

0x50fd88 LDR	X10, [X15, #24]    [1]

0x50fd8c ADD	X13, X7, X9,LSL #4

0x50fd90 ADD	X3, X16, X5,LSL #3

0x50fd94 LDP	D30, D29, [X23, #8]    [10]

0x50fd98 ADD	X17, X16, X5,LSL #4

0x50fd9c CMP	W11, #0

0x50fda0 B.LE	50fd0c

0x50feec BLR	X2

0x50fef0 LDR	W11, [X19, #16]    [11]

0x50fef4 B	50fd58

/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/unique_ptr.h: 193 - 193

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

193:       pointer    _M_ptr() const noexcept { return std::get<0>(_M_t); }

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/RealSpacePositionsOMPTarget.h: 163 - 163

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

163:   const PosVectorSoa& getAllParticlePos() const override { return RSoA_hostview; }

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Utilities/NewTimer.h: 249 - 249

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

249:   ~ScopeGuard() { timer.stop(); }

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/Lattice/ParticleBConds3DSoa.h: 222 - 234

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

222:     const T x0 = pos[0];

223:     const T y0 = pos[1];

224:     const T z0 = pos[2];

[...]

234: #pragma omp simd aligned(temp_r, px, py, pz, dx, dy, dz: QMC_SIMD_ALIGNMENT)

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h: 271 - 273

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

271:   inline T* restrict data(size_t i) { return myData + i * nGhosts; }

272:   ///return the const pointer of the i-th components

273:   inline const T* restrict data(size_t i) const { return myData + i * nGhosts; }

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/OhmmsVector.h: 229 - 248

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

229:     return X[i];

[...]

248:   inline pointer data() { return X; }

/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_vector.h: 1131 - 1131

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

1131: 	return *(this->_M_impl._M_start + __n);

/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableABOMPTarget.h: 366 - 368

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

366:       DTD_BConds<T, D, SC>::computeDistances(P.R[iat], origin_.getCoordinates().getAllParticlePos(),

367:                                              distances_[iat].data(), displacements_[iat], 0, num_sources_);

368:   }

Coverage (%)	Name	Source Location	Module

Path /

Metric	Value
CQA speedup if no scalar integer	6.67
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	6.67
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.30
Bottlenecks	P10, P11, P12,
Function	qmcplusplus::SoaDistanceTableABOMPTarget::move(qmcplusplus::ParticleSet const&, qmcplusplus::TinyVector const&, int, bool)
Source	unique_ptr.h:193-193,RealSpacePositionsOMPTarget.h:163-163,NewTimer.h:249-249,ParticleBConds3DSoa.h:222-224,ParticleBConds3DSoa.h:234-234,VectorSoAContainer.h:271-273,OhmmsVector.h:229-229,OhmmsVector.h:248-248,stl_vector.h:1131-1131,SoaDistanceTableABOMPTarget.h:366-368
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	6.67
CQA cycles if no scalar integer	1.00
CQA cycles if FP arith vectorized	6.67
CQA cycles if fully vectorized	1.00
Front-end cycles	5.13
DIV/SQRT cycles	3.00
P0 cycles	3.00
P1 cycles	4.25
P2 cycles	4.25
P3 cycles	4.25
P4 cycles	4.25
P5 cycles	0.00
P6 cycles	0.00
P7 cycles	0.00
P8 cycles	0.00
P9 cycles	6.67
P10 cycles	6.67
P11 cycles	6.67
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	0.00
Inter-iter dependencies cycles	NA
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	41.00
Nb uops	41.00
Nb loads	NA
Nb stores	0.00
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	29.40
Bytes prefetched	0.00
Bytes loaded	196.00
Bytes stored	0.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	11.11
Vectorization ratio load	50.00
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	0.00
Vectorization ratio fma	0.00
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	25.00
Vector-efficiency ratio load	37.50
Vector-efficiency ratio store	NA
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	25.00
Vector-efficiency ratio fma	12.50
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	18.75

Metric	Value
CQA speedup if no scalar integer	6.67
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	6.67
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.30
Bottlenecks	P10, P11, P12,
Function	qmcplusplus::SoaDistanceTableABOMPTarget::move(qmcplusplus::ParticleSet const&, qmcplusplus::TinyVector const&, int, bool)
Source	unique_ptr.h:193-193,RealSpacePositionsOMPTarget.h:163-163,NewTimer.h:249-249,ParticleBConds3DSoa.h:222-224,ParticleBConds3DSoa.h:234-234,VectorSoAContainer.h:271-273,OhmmsVector.h:229-229,OhmmsVector.h:248-248,stl_vector.h:1131-1131,SoaDistanceTableABOMPTarget.h:366-368
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	6.67
CQA cycles if no scalar integer	1.00
CQA cycles if FP arith vectorized	6.67
CQA cycles if fully vectorized	1.00
Front-end cycles	5.13
DIV/SQRT cycles	3.00
P0 cycles	3.00
P1 cycles	4.25
P2 cycles	4.25
P3 cycles	4.25
P4 cycles	4.25
P5 cycles	0.00
P6 cycles	0.00
P7 cycles	0.00
P8 cycles	0.00
P9 cycles	6.67
P10 cycles	6.67
P11 cycles	6.67
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	0.00
Inter-iter dependencies cycles	NA
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	41.00
Nb uops	41.00
Nb loads	NA
Nb stores	0.00
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	29.40
Bytes prefetched	0.00
Bytes loaded	196.00
Bytes stored	0.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	11.11
Vectorization ratio load	50.00
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	0.00
Vectorization ratio fma	0.00
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	25.00
Vector-efficiency ratio load	37.50
Vector-efficiency ratio store	NA
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	25.00
Vector-efficiency ratio fma	12.50
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	18.75

Path /

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

Function	qmcplusplus::SoaDistanceTableABOMPTarget::move(qmcplusplus::ParticleSet const&, qmcplusplus::TinyVector const&, int, bool)
Source file and lines	VectorSoAContainer.h:271-273
Module	exec

The loop is defined in:

/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/unique_ptr.h:193
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/RealSpacePositionsOMPTarget.h:163
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Utilities/NewTimer.h:249
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/Lattice/ParticleBConds3DSoa.h:222-224,234
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h:271-273
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/OhmmsVector.h:229,248
/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_vector.h:1131
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableABOMPTarget.h:366-368

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

Ignoring paths for analysis
Failed to get the number of paths
RecMII not computed since number of paths is unknown or > max_paths

gain
potential
hint
expert

Code clean check

Detected a slowdown caused by scalar integer instructions (typically used for address computation). By removing them, you can lower the cost of an iteration from 6.67 to 1.00 cycles (6.67x speedup).

Workaround

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

Vectorization

Your loop is not vectorized. Only 25% 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 6.67 to 1.00 cycles (6.67x speedup).

Details

11% of VPU instructions are used in vector version (process two or more data elements in vector registers):

50% of VPU loads are used in vector version.
0% of VPU addition or subtraction instructions are used in vector version.
0% of VPU fused multiply-add instructions are used in vector version.
0% 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)

Expensive FP math instructions/calls

Detected performance impact from expensive FP math instructions/calls. By removing/reexpressing them, you can lower the cost of an iteration from 6.67 to 1.00 cycles (6.67x speedup).

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 196 bytes.

General properties

nb instructions	41
loop length	164
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

5.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	3.00	3.00	4.25	4.25	4.25	4.25	0.00	0.00	0.00	0.00	6.67	6.67	6.67	0.00	0.00
cycles	3.00	3.00	4.25	4.25	4.25	4.25	0.00	0.00	0.00	0.00	6.67	6.67	6.67	0.00	0.00

Execution ports to units layout:

P0:
P1:
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

Cycles summary

Front-end	5.13
Overall L1	6.67

Vectorization ratios

all	11%
load	50%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	0%
fma	0%
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	0%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 6.67 cycles. At this rate:

30% of peak load performance is reached (29.40 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 50fd2c

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P10	P11	P12	Latency	Recip. throughput
ORR X0, XZR, X21	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
BL 4c4660 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE4stopEv>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	1	0.50
LDR X25, [SP, #64]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDP X19, X20, [SP, #16]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
LDP X21, X22, [SP, #32]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
LDP X23, X24, [SP, #48]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
LDP X29, X30, [SP], #80	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
RET	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50
LDR X0, [X19, #8]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
MOVZ W23, #24	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X24, [X22, #64]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
SMADDL X22, W20, W23, XZR	1	0	0	0	0	1	0	0	0	0	2	1
LDR X0, [X0, #632]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ADD X23, X24, X22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X1, [X0]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR X2, [X1, #72]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
CMP X2, X25	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.33
B.NE 50feec <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE4moveERKNS_11ParticleSetERKNS_10TinyVectorIdLj3EEEib+0x3ac>	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50
ADD X0, X0, #64	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X14, [X19, #96]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
MOVZ W25, #40	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
SMADDL X20, W20, W25, XZR	1	0	0	0	0	1	0	0	0	0	2	1
LDR X12, [X19, #72]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR X9, [X0, #8]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ADD X6, X14, X20	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X5, [X6, #8]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ADD X15, X12, X20	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X7, [X0, #24]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR X16, [X6, #24]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR D31, [X24, X22]	1	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33
ADD X8, X7, X9,LSL #3	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X10, [X15, #24]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ADD X13, X7, X9,LSL #4	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
ADD X3, X16, X5,LSL #3	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDP D30, D29, [X23, #8]	1	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33
ADD X17, X16, X5,LSL #4	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
CMP W11, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.33
B.LE 50fd0c <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE4moveERKNS_11ParticleSetERKNS_10TinyVectorIdLj3EEEib+0x1cc>	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50
BLR X2	1	0.50	0.50	0.50	0.50	0	0	0	0	0	1	0.50
LDR W11, [X19, #16]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
B 50fd58 <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE4moveERKNS_11ParticleSetERKNS_10TinyVectorIdLj3EEEib+0x218>	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50

Function	qmcplusplus::SoaDistanceTableABOMPTarget::move(qmcplusplus::ParticleSet const&, qmcplusplus::TinyVector const&, int, bool)
Source file and lines	VectorSoAContainer.h:271-273
Module	exec

The loop is defined in:

/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/unique_ptr.h:193
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/RealSpacePositionsOMPTarget.h:163
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Utilities/NewTimer.h:249
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/Lattice/ParticleBConds3DSoa.h:222-224,234
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h:271-273
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/OhmmsVector.h:229,248
/home/hbollore/gnu/unknow_version/include/c++/14.0.1/bits/stl_vector.h:1131
/home/hbollore/qaas-runs/171-275-7410/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableABOMPTarget.h:366-368

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

Ignoring paths for analysis
Failed to get the number of paths
RecMII not computed since number of paths is unknown or > max_paths

gain
potential
hint
expert

Code clean check

Detected a slowdown caused by scalar integer instructions (typically used for address computation). By removing them, you can lower the cost of an iteration from 6.67 to 1.00 cycles (6.67x speedup).

Workaround

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

Vectorization

Details

11% of VPU instructions are used in vector version (process two or more data elements in vector registers):

50% of VPU loads are used in vector version.
0% of VPU addition or subtraction instructions are used in vector version.
0% of VPU fused multiply-add instructions are used in vector version.
0% 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)

Expensive FP math instructions/calls

Detected performance impact from expensive FP math instructions/calls. By removing/reexpressing them, you can lower the cost of an iteration from 6.67 to 1.00 cycles (6.67x speedup).

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 196 bytes.

General properties

nb instructions	41
loop length	164
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

5.13 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	3.00	3.00	4.25	4.25	4.25	4.25	0.00	0.00	0.00	0.00	6.67	6.67	6.67	0.00	0.00
cycles	3.00	3.00	4.25	4.25	4.25	4.25	0.00	0.00	0.00	0.00	6.67	6.67	6.67	0.00	0.00

Execution ports to units layout:

P0:
P1:
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

Cycles summary

Front-end	5.13
Overall L1	6.67

Vectorization ratios

all	11%
load	50%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	0%
fma	0%
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	0%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 6.67 cycles. At this rate:

30% of peak load performance is reached (29.40 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 50fd2c

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P10	P11	P12	Latency	Recip. throughput
ORR X0, XZR, X21	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
BL 4c4660 <_ZN11qmcplusplus9TimerTypeINSt6chrono3_V212system_clockEE4stopEv>	1	0.50	0.50	0.50	0.50	0	0	0	0	0	1	0.50
LDR X25, [SP, #64]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDP X19, X20, [SP, #16]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
LDP X21, X22, [SP, #32]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
LDP X23, X24, [SP, #48]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
LDP X29, X30, [SP], #80	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
RET	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50
LDR X0, [X19, #8]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
MOVZ W23, #24	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X24, [X22, #64]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
SMADDL X22, W20, W23, XZR	1	0	0	0	0	1	0	0	0	0	2	1
LDR X0, [X0, #632]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ADD X23, X24, X22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X1, [X0]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR X2, [X1, #72]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
CMP X2, X25	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.33
B.NE 50feec <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE4moveERKNS_11ParticleSetERKNS_10TinyVectorIdLj3EEEib+0x3ac>	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50
ADD X0, X0, #64	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X14, [X19, #96]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
MOVZ W25, #40	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
SMADDL X20, W20, W25, XZR	1	0	0	0	0	1	0	0	0	0	2	1
LDR X12, [X19, #72]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR X9, [X0, #8]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ADD X6, X14, X20	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X5, [X6, #8]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ADD X15, X12, X20	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X7, [X0, #24]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR X16, [X6, #24]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR D31, [X24, X22]	1	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33
ADD X8, X7, X9,LSL #3	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X10, [X15, #24]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ADD X13, X7, X9,LSL #4	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
ADD X3, X16, X5,LSL #3	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDP D30, D29, [X23, #8]	1	0	0	0	0	0	0	0.33	0.33	0.33	6	0.33
ADD X17, X16, X5,LSL #4	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
CMP W11, #0	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.33
B.LE 50fd0c <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE4moveERKNS_11ParticleSetERKNS_10TinyVectorIdLj3EEEib+0x1cc>	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50
BLR X2	1	0.50	0.50	0.50	0.50	0	0	0	0	0	1	0.50
LDR W11, [X19, #16]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
B 50fd58 <_ZN11qmcplusplus27SoaDistanceTableABOMPTargetIdLj3ELi40EE4moveERKNS_11ParticleSetERKNS_10TinyVectorIdLj3EEEib+0x218>	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50

Report Configuration

Code clean check

Workaround

Vectorization

Details

Workaround

Expensive FP math instructions/calls

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

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Cycles and memory resources usage

ASM code

Code clean check

Workaround

Vectorization

Details

Workaround

Expensive FP math instructions/calls

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

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Cycles and memory resources usage

ASM code