OV - exec - Loop 1087

0x463100 LDR	X0, [X8, X19,LSL #3]    [4]

0x463104 ORR	W3, WZR, W22

0x463108 ORR	X2, XZR, X23

0x46310c ORR	X1, XZR, X21

0x463110 LDR	X4, [X0]    [2]

0x463114 LDR	X7, [X4, #32]    [3]

0x463118 BLR	X7

0x46311c LDP	X8, X9, [X20]    [1]

0x463120 ADD	X19, X19, #1

0x463124 SUB	X10, X9, X8

0x463128 CMP	X19, X10,ASR #3

0x46312c B.CC	463100

/home/hbollore/qaas/qaas-runs/169-816-8620/intel/miniqmc/build/miniqmc/src/Particle/ParticleSet.cpp: 290 - 291

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

290:   for (int i = 0; i < DistTables.size(); ++i)

291:     DistTables[i]->move(*this, activePos, iat);

/usr/include/c++/11/bits/stl_vector.h: 919 - 919

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

919:       { return size_type(this->_M_impl._M_finish - this->_M_impl._M_start); }

Coverage (%)	Name	Source Location	Module
►100.00+	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

Path /

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	6.22
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.17
Bottlenecks	P2, P3, P4, P5,
Function	qmcplusplus::ParticleSet::makeMove(int, qmcplusplus::TinyVector const&)
Source	ParticleSet.cpp:290-291,stl_vector.h:919-919
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	1.75
CQA cycles if no scalar integer	1.75
CQA cycles if FP arith vectorized	1.75
CQA cycles if fully vectorized	0.28
Front-end cycles	1.50
DIV/SQRT cycles	1.00
P0 cycles	1.00
P1 cycles	1.75
P2 cycles	1.75
P3 cycles	1.75
P4 cycles	1.75
P5 cycles	0.00
P6 cycles	0.00
P7 cycles	0.00
P8 cycles	0.00
P9 cycles	1.33
P10 cycles	1.33
P11 cycles	1.33
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	12.00
Nb uops	12.00
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	22.86
Bytes prefetched	0.00
Bytes loaded	40.00
Bytes stored	0.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	0.00
Vectorization ratio load	NA
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	20.83
Vector-efficiency ratio load	NA
Vector-efficiency ratio store	NA
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	20.83

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	6.22
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.17
Bottlenecks	P2, P3, P4, P5,
Function	qmcplusplus::ParticleSet::makeMove(int, qmcplusplus::TinyVector const&)
Source	ParticleSet.cpp:290-291,stl_vector.h:919-919
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	1.75
CQA cycles if no scalar integer	1.75
CQA cycles if FP arith vectorized	1.75
CQA cycles if fully vectorized	0.28
Front-end cycles	1.50
DIV/SQRT cycles	1.00
P0 cycles	1.00
P1 cycles	1.75
P2 cycles	1.75
P3 cycles	1.75
P4 cycles	1.75
P5 cycles	0.00
P6 cycles	0.00
P7 cycles	0.00
P8 cycles	0.00
P9 cycles	1.33
P10 cycles	1.33
P11 cycles	1.33
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	12.00
Nb uops	12.00
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	22.86
Bytes prefetched	0.00
Bytes loaded	40.00
Bytes stored	0.00
Stride 0	NA
Stride 1	NA
Stride n	NA
Stride unknown	NA
Stride indirect	NA
Vectorization ratio all	0.00
Vectorization ratio load	NA
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	0.00
Vector-efficiency ratio all	20.83
Vector-efficiency ratio load	NA
Vector-efficiency ratio store	NA
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	20.83

Path /

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

Function	qmcplusplus::ParticleSet::makeMove(int, qmcplusplus::TinyVector const&)
Source file and lines	ParticleSet.cpp:290-291
Module	exec

The loop is defined in:

/home/hbollore/qaas/qaas-runs/169-816-8620/intel/miniqmc/build/miniqmc/src/Particle/ParticleSet.cpp:290-291
/usr/include/c++/11/bits/stl_vector.h:919

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

gain
potential
hint
expert

Vectorization

Your loop is not vectorized. Only 20% of vector register length is used (average across all VPU instructions). By vectorizing your loop, you can lower the cost of an iteration from 1.75 to 0.28 cycles (6.22x speedup).

Details

All VPU 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:
- recompile with ftree-vectorize (included in O3) to enable loop vectorization (including FP reductions).
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 1.75 to 1.00 cycles (1.75x 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 40 bytes.

General properties

nb instructions	12
loop length	48
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

1.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	1.00	1.00	1.75	1.75	1.75	1.75	0.00	0.00	0.00	0.00	1.33	1.33	1.33	0.00	0.00
cycles	1.00	1.00	1.75	1.75	1.75	1.75	0.00	0.00	0.00	0.00	1.33	1.33	1.33	0.00	0.00

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

Cycles summary

Front-end	1.50
Data deps.	1.00
Overall L1	1.75

Vectorization ratios

all	0%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
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	0%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 1.75 cycles. At this rate:

23% of peak load performance is reached (22.86 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 463100

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P10	P11	P12	Latency	Recip. throughput
LDR X0, [X8, X19,LSL #3]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ORR W3, WZR, W22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
ORR X2, XZR, X23	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
ORR X1, XZR, X21	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X4, [X0]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR X7, [X4, #32]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
BLR X7	1	0.50	0.50	0.50	0.50	0	0	0	0	0	1	0.50
LDP X8, X9, [X20]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
ADD X19, X19, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
SUB X10, X9, X8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
CMP X19, X10,ASR #3	1	0	0	0	0	0.50	0.50	0	0	0	2	0.50
B.CC 463100 <_ZN11qmcplusplus11ParticleSet8makeMoveEiRKNS_10TinyVectorIdLj3EEE+0x90>	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50

Function	qmcplusplus::ParticleSet::makeMove(int, qmcplusplus::TinyVector const&)
Source file and lines	ParticleSet.cpp:290-291
Module	exec

The loop is defined in:

/home/hbollore/qaas/qaas-runs/169-816-8620/intel/miniqmc/build/miniqmc/src/Particle/ParticleSet.cpp:290-291
/usr/include/c++/11/bits/stl_vector.h:919

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

gain
potential
hint
expert

Vectorization

Your loop is not vectorized. Only 20% of vector register length is used (average across all VPU instructions). By vectorizing your loop, you can lower the cost of an iteration from 1.75 to 0.28 cycles (6.22x speedup).

Details

All VPU 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:
- recompile with ftree-vectorize (included in O3) to enable loop vectorization (including FP reductions).
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 1.75 to 1.00 cycles (1.75x 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 40 bytes.

General properties

nb instructions	12
loop length	48
nb stack references	0

Front-end

FIT IN UOP CACHE

front end

1.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14
uops	1.00	1.00	1.75	1.75	1.75	1.75	0.00	0.00	0.00	0.00	1.33	1.33	1.33	0.00	0.00
cycles	1.00	1.00	1.75	1.75	1.75	1.75	0.00	0.00	0.00	0.00	1.33	1.33	1.33	0.00	0.00

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

Cycles summary

Front-end	1.50
Data deps.	1.00
Overall L1	1.75

Vectorization ratios

all	0%
load	NA (no load vectorizable/vectorized instructions)
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
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	0%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 1.75 cycles. At this rate:

23% of peak load performance is reached (22.86 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))

ASM code

In the binary file, the address of the loop is: 463100

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P10	P11	P12	Latency	Recip. throughput
LDR X0, [X8, X19,LSL #3]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
ORR W3, WZR, W22	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
ORR X2, XZR, X23	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
ORR X1, XZR, X21	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
LDR X4, [X0]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
LDR X7, [X4, #32]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	0.33
BLR X7	1	0.50	0.50	0.50	0.50	0	0	0	0	0	1	0.50
LDP X8, X9, [X20]	1	0	0	0	0	0	0	0.33	0.33	0.33	4	1
ADD X19, X19, #1	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
SUB X10, X9, X8	1	0	0	0.25	0.25	0.25	0.25	0	0	0	1	0.25
CMP X19, X10,ASR #3	1	0	0	0	0	0.50	0.50	0	0	0	2	0.50
B.CC 463100 <_ZN11qmcplusplus11ParticleSet8makeMoveEiRKNS_10TinyVectorIdLj3EEE+0x90>	1	0.50	0.50	0	0	0	0	0	0	0	1	0.50

Report Configuration

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

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