OV - - Loop 2304

	Loop Id: 2304	Module: exec	Source: SoaDistanceTableAAOMPTarget.h:440-442 [...]	Coverage: 0.81%

0x554ec0 MOV	0x30(%RDI,%R12,1),%R14    [6]

0x554ec5 MOV	0x40(%R15,%R12,1),%RCX    [18]

0x554eca MOV	0x40(%RDI,%R12,1),%RDX    [6]

0x554ecf VMOVSD	(%R8,%RAX,8),%XMM12    [13]

0x554ed5 SAL	$0x3,%R14

0x554ed9 VMOVSD	%XMM12,(%RCX,%RBX,1)    [17]

0x554ede VMOVSD	(%R11,%RAX,8),%XMM13    [14]

0x554ee4 VMOVSD	(%R10,%RAX,8),%XMM1    [10]

0x554eea VXORPD	%XMM11,%XMM13,%XMM14

0x554eef VXORPD	%XMM11,%XMM1,%XMM2

0x554ef4 VMOVSD	(%R9,%RAX,8),%XMM15    [8]

0x554efa VXORPD	%XMM11,%XMM15,%XMM0

0x554eff LEA	(%R14,%RBX,1),%R13

0x554f03 VMOVSD	%XMM2,(%RDX,%RBX,1)    [15]

0x554f08 LEA	0x1(%RAX),%RCX

0x554f0c ADD	$0x2,%RAX

0x554f10 VMOVSD	%XMM14,(%RDX,%R13,1)    [11]

0x554f16 ADD	%R14,%RDX

0x554f19 MOV	0x68(%R15,%R12,1),%R14    [18]

0x554f1e VMOVSD	%XMM0,(%RDX,%R13,1)    [2]

0x554f24 MOV	0x58(%RDI,%R12,1),%R13    [6]

0x554f29 MOV	0x68(%RDI,%R12,1),%RDX    [6]

0x554f2e VMOVSD	(%R8,%RCX,8),%XMM3    [1]

0x554f34 ADD	$0x50,%R12

0x554f38 VMOVSD	%XMM3,(%R14,%RBX,1)    [5]

0x554f3e VMOVSD	(%R11,%RCX,8),%XMM4    [12]

0x554f44 VMOVSD	(%R9,%RCX,8),%XMM6    [3]

0x554f4a VMOVSD	(%R10,%RCX,8),%XMM8    [16]

0x554f50 SAL	$0x3,%R13

0x554f54 VXORPD	%XMM11,%XMM4,%XMM5

0x554f59 VXORPD	%XMM11,%XMM8,%XMM9

0x554f5e VXORPD	%XMM11,%XMM6,%XMM7

0x554f63 LEA	(%R13,%RBX,1),%RCX

0x554f68 VMOVSD	%XMM9,(%RDX,%RBX,1)    [9]

0x554f6d VMOVSD	%XMM5,(%RDX,%RCX,1)    [7]

0x554f72 ADD	%R13,%RDX

0x554f75 VMOVSD	%XMM7,(%RDX,%RCX,1)    [4]

0x554f7a CMP	%RSI,%RAX

0x554f7d JNE	554ec0

/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h: 243 - 263

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

243:       for (size_t i = 0; i < D; ++i)

244:         *(_base + M * i) = rhs[i];

[...]

263:   inline Accessor operator()(size_t i) { return Accessor(myData + i, nGhosts); }

/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/TinyVector.h: 182 - 182

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

182:       inverse[d] = -X[d];

/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableAAOMPTarget.h: 440 - 442

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

440:     for (size_t i = iat + 1; i < num_targets_; ++i)

441:     {

442:       distances_[i][iat]     = temp_r_[i];

Path /

Metric	Value
CQA speedup if no scalar integer	1.38
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	4.89
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.16
Bottlenecks	P5, P6, P7,
Function	_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi
Source	VectorSoAContainer.h:243-244,VectorSoAContainer.h:263-263,TinyVector.h:182-182,SoaDistanceTableAAOMPTarget.h:440-442
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	7.33
CQA cycles if no scalar integer	5.33
CQA cycles if FP arith vectorized	7.33
CQA cycles if fully vectorized	1.50
Front-end cycles	6.33
DIV/SQRT cycles	2.50
P0 cycles	2.50
P1 cycles	2.25
P2 cycles	2.25
P3 cycles	0.50
P4 cycles	7.33
P5 cycles	7.33
P6 cycles	7.33
P7 cycles	1.50
P8 cycles	1.50
P9 cycles	1.50
P10 cycles	1.50
P11 cycles	4.00
P12 cycles	4.00
P13 cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	39.00
Nb uops	38.00
Nb loads	14.00
Nb stores	8.00
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	24.00
Bytes prefetched	0.00
Bytes loaded	112.00
Bytes stored	64.00
Stride 0	0.00
Stride 1	0.00
Stride n	6.00
Stride unknown	9.00
Stride indirect	0.00
Vectorization ratio all	27.27
Vectorization ratio load	0.00
Vectorization ratio store	0.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	100.00
Vector-efficiency ratio all	15.91
Vector-efficiency ratio load	12.50
Vector-efficiency ratio store	12.50
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	25.00

Metric	Value
CQA speedup if no scalar integer	1.38
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	4.89
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.16
Bottlenecks	P5, P6, P7,
Function	_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi
Source	VectorSoAContainer.h:243-244,VectorSoAContainer.h:263-263,TinyVector.h:182-182,SoaDistanceTableAAOMPTarget.h:440-442
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	7.33
CQA cycles if no scalar integer	5.33
CQA cycles if FP arith vectorized	7.33
CQA cycles if fully vectorized	1.50
Front-end cycles	6.33
DIV/SQRT cycles	2.50
P0 cycles	2.50
P1 cycles	2.25
P2 cycles	2.25
P3 cycles	0.50
P4 cycles	7.33
P5 cycles	7.33
P6 cycles	7.33
P7 cycles	1.50
P8 cycles	1.50
P9 cycles	1.50
P10 cycles	1.50
P11 cycles	4.00
P12 cycles	4.00
P13 cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	39.00
Nb uops	38.00
Nb loads	14.00
Nb stores	8.00
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	24.00
Bytes prefetched	0.00
Bytes loaded	112.00
Bytes stored	64.00
Stride 0	0.00
Stride 1	0.00
Stride n	6.00
Stride unknown	9.00
Stride indirect	0.00
Vectorization ratio all	27.27
Vectorization ratio load	0.00
Vectorization ratio store	0.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	100.00
Vector-efficiency ratio all	15.91
Vector-efficiency ratio load	12.50
Vector-efficiency ratio store	12.50
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	25.00

Path /

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

Function	_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi
Source file and lines	SoaDistanceTableAAOMPTarget.h:440-442
Module	exec

The loop is defined in:

/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h:243-244,263
/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/TinyVector.h:182
/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableAAOMPTarget.h:440-442

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

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 7.33 to 5.33 cycles (1.37x speedup).

Workaround

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

Vectorization

Your loop is probably not vectorized. Only 15% of vector register length is used (average across all SSE/AVX instructions). By vectorizing your loop, you can lower the cost of an iteration from 7.33 to 1.50 cycles (4.89x speedup).

Details

Store and arithmetical 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:

reading data from caches/RAM (load units are a bottleneck)
writing data to caches/RAM (the store unit is a bottleneck)

By removing all these bottlenecks, you can lower the cost of an iteration from 7.33 to 6.33 cycles (1.16x speedup).

Workaround

Read less array elements
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 unknown stride: 9 occurrence(s)
Constant non-unit stride: 6 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

No instructions are processing arithmetic or math operations on FP elements. This loop is probably writing/copying data or processing integer elements.

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 112 bytes. The binary loop is storing 64 bytes.

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	39
nb uops	38
loop length	195
used x86 registers	14
used mmx registers	0
used xmm registers	15
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	6.33 cycles
front end	6.33 cycles

Back-end

	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5
uops	2.50	2.50	2.25	2.25	0.50	7.33	7.33	7.33	1.50	1.50	1.50	1.50	4.00	4.00
cycles	2.50	2.50	2.25	2.25	0.50	7.33	7.33	7.33	1.50	1.50	1.50	1.50	4.00	4.00

Execution ports to units layout:

ALU0/BRU0: ALU
ALU1: ALU
ALU2: ALU
ALU3: ALU
BRU1:
AGU0 (256 bits): store address, load
AGU1 (256 bits): store address, load
AGU2 (256 bits): store address, load
FP0 (256 bits): VPU, DIV/SQRT
FP1 (256 bits): VPU, DIV/SQRT
FP2 (256 bits): VPU
FP3 (256 bits): VPU
FP4 (256 bits): FP store data
FP5 (256 bits): FP store data

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

Cycles summary

Front-end	6.33
Dispatch	7.33
Data deps.	1.00
Overall L1	7.33

Vectorization ratios

all	27%
load	0%
store	0%
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	100%

Vector efficiency ratios

all	15%
load	12%
store	12%
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	25%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 7.33 cycles. At this rate:

15% of peak load performance is reached (15.27 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
13% of peak store performance is reached (8.73 out of 64.00 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: 554ec0

Instruction	Nb FU	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5	Latency	Recip. throughput
MOV 0x30(%RDI,%R12,1),%R14	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
MOV 0x40(%R15,%R12,1),%RCX	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
MOV 0x40(%RDI,%R12,1),%RDX	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
VMOVSD (%R8,%RAX,8),%XMM12	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
SAL $0x3,%R14	1	0	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
VMOVSD %XMM12,(%RCX,%RBX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
VMOVSD (%R11,%RAX,8),%XMM13	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VMOVSD (%R10,%RAX,8),%XMM1	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VXORPD %XMM11,%XMM13,%XMM14	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
VXORPD %XMM11,%XMM1,%XMM2	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
VMOVSD (%R9,%RAX,8),%XMM15	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VXORPD %XMM11,%XMM15,%XMM0	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
LEA (%R14,%RBX,1),%R13	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM2,(%RDX,%RBX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
LEA 0x1(%RAX),%RCX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
ADD $0x2,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM14,(%RDX,%R13,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
ADD %R14,%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
MOV 0x68(%R15,%R12,1),%R14	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
VMOVSD %XMM0,(%RDX,%R13,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
MOV 0x58(%RDI,%R12,1),%R13	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
MOV 0x68(%RDI,%R12,1),%RDX	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
VMOVSD (%R8,%RCX,8),%XMM3	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
ADD $0x50,%R12	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM3,(%R14,%RBX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
VMOVSD (%R11,%RCX,8),%XMM4	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VMOVSD (%R9,%RCX,8),%XMM6	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VMOVSD (%R10,%RCX,8),%XMM8	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
SAL $0x3,%R13	1	0	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
VXORPD %XMM11,%XMM4,%XMM5	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
VXORPD %XMM11,%XMM8,%XMM9	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
VXORPD %XMM11,%XMM6,%XMM7	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
LEA (%R13,%RBX,1),%RCX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM9,(%RDX,%RBX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
VMOVSD %XMM5,(%RDX,%RCX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
ADD %R13,%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM7,(%RDX,%RCX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
CMP %RSI,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
JNE 554ec0 <_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi+0x170>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1

Function	_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi
Source file and lines	SoaDistanceTableAAOMPTarget.h:440-442
Module	exec

The loop is defined in:

/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/VectorSoAContainer.h:243-244,263
/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/Numerics/OhmmsPETE/TinyVector.h:182
/beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/build/miniqmc/src/Particle/SoaDistanceTableAAOMPTarget.h:440-442

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

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 7.33 to 5.33 cycles (1.37x speedup).

Workaround

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

Vectorization

Details

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:

reading data from caches/RAM (load units are a bottleneck)
writing data to caches/RAM (the store unit is a bottleneck)

By removing all these bottlenecks, you can lower the cost of an iteration from 7.33 to 6.33 cycles (1.16x speedup).

Workaround

Read less array elements
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 unknown stride: 9 occurrence(s)
Constant non-unit stride: 6 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

No instructions are processing arithmetic or math operations on FP elements. This loop is probably writing/copying data or processing integer elements.

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 112 bytes. The binary loop is storing 64 bytes.

Unroll opportunity

Loop is data access bound.

Workaround

General properties

nb instructions	39
nb uops	38
loop length	195
used x86 registers	14
used mmx registers	0
used xmm registers	15
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	6.33 cycles
front end	6.33 cycles

Back-end

	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5
uops	2.50	2.50	2.25	2.25	0.50	7.33	7.33	7.33	1.50	1.50	1.50	1.50	4.00	4.00
cycles	2.50	2.50	2.25	2.25	0.50	7.33	7.33	7.33	1.50	1.50	1.50	1.50	4.00	4.00

Execution ports to units layout:

ALU0/BRU0: ALU
ALU1: ALU
ALU2: ALU
ALU3: ALU
BRU1:
AGU0 (256 bits): store address, load
AGU1 (256 bits): store address, load
AGU2 (256 bits): store address, load
FP0 (256 bits): VPU, DIV/SQRT
FP1 (256 bits): VPU, DIV/SQRT
FP2 (256 bits): VPU
FP3 (256 bits): VPU
FP4 (256 bits): FP store data
FP5 (256 bits): FP store data

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

Cycles summary

Front-end	6.33
Dispatch	7.33
Data deps.	1.00
Overall L1	7.33

Vectorization ratios

all	27%
load	0%
store	0%
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	100%

Vector efficiency ratios

all	15%
load	12%
store	12%
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	25%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 7.33 cycles. At this rate:

15% of peak load performance is reached (15.27 out of 96.00 bytes loaded per cycle (GB/s @ 1GHz))
13% of peak store performance is reached (8.73 out of 64.00 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: 554ec0

Instruction	Nb FU	ALU0/BRU0	ALU1	ALU2	ALU3	BRU1	AGU0	AGU1	AGU2	FP0	FP1	FP2	FP3	FP4	FP5	Latency	Recip. throughput
MOV 0x30(%RDI,%R12,1),%R14	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
MOV 0x40(%R15,%R12,1),%RCX	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
MOV 0x40(%RDI,%R12,1),%RDX	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
VMOVSD (%R8,%RAX,8),%XMM12	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
SAL $0x3,%R14	1	0	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
VMOVSD %XMM12,(%RCX,%RBX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
VMOVSD (%R11,%RAX,8),%XMM13	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VMOVSD (%R10,%RAX,8),%XMM1	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VXORPD %XMM11,%XMM13,%XMM14	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
VXORPD %XMM11,%XMM1,%XMM2	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
VMOVSD (%R9,%RAX,8),%XMM15	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VXORPD %XMM11,%XMM15,%XMM0	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
LEA (%R14,%RBX,1),%R13	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM2,(%RDX,%RBX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
LEA 0x1(%RAX),%RCX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
ADD $0x2,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM14,(%RDX,%R13,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
ADD %R14,%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
MOV 0x68(%R15,%R12,1),%R14	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
VMOVSD %XMM0,(%RDX,%R13,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
MOV 0x58(%RDI,%R12,1),%R13	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
MOV 0x68(%RDI,%R12,1),%RDX	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	3	0.33
VMOVSD (%R8,%RCX,8),%XMM3	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
ADD $0x50,%R12	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM3,(%R14,%RBX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
VMOVSD (%R11,%RCX,8),%XMM4	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VMOVSD (%R9,%RCX,8),%XMM6	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
VMOVSD (%R10,%RCX,8),%XMM8	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	1	0.50
SAL $0x3,%R13	1	0	0.50	0.50	0	0	0	0	0	0	0	0	0	0	0	1	0.50
VXORPD %XMM11,%XMM4,%XMM5	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
VXORPD %XMM11,%XMM8,%XMM9	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
VXORPD %XMM11,%XMM6,%XMM7	1	0	0	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	1	0.25
LEA (%R13,%RBX,1),%RCX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM9,(%RDX,%RBX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
VMOVSD %XMM5,(%RDX,%RCX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
ADD %R13,%RDX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
VMOVSD %XMM7,(%RDX,%RCX,1)	1	0	0	0	0	0	0.33	0.33	0.33	0	0	0	0	0.50	0.50	1	1
CMP %RSI,%RAX	1	0.25	0.25	0.25	0.25	0	0	0	0	0	0	0	0	0	0	1	0.25
JNE 554ec0 <_ZN11qmcplusplus27SoaDistanceTableAAOMPTargetIdLj3ELi40EE6updateEi+0x170>	1	0.50	0	0	0	0.50	0	0	0	0	0	0	0	0	0	1	0.50-1

▶Scalability Runs Description

Run 2x1	Number processes: 2Number nodes: 1Number processes per node: 2Run Command: <executable> -g "4 2 2" -bMPI Command: mpirun -np <number_processes>Dataset: Run Directory: /beegfs/hackathon/users/eoseret/qaas_runs/170-855-3059/intel/miniqmc/run/oneview_runs/compilers/gcc_2/oneview_run_1708580749I_MPI_PIN_DOMAIN: auto:scatterOMP_PLACES: threadsOMP_PROC_BIND: spreadOMP_NUM_THREADS: 1
Run 2x2	OMP_NUM_THREADS: 2I_MPI_PIN_DOMAIN: auto:scatterOMP_PLACES: threadsOMP_PROC_BIND: spread
Run 2x4	OMP_NUM_THREADS: 4I_MPI_PIN_DOMAIN: auto:scatterOMP_PLACES: threadsOMP_PROC_BIND: spread
Run 2x8	OMP_NUM_THREADS: 8I_MPI_PIN_DOMAIN: auto:scatterOMP_PLACES: threadsOMP_PROC_BIND: spread
Run 2x16	OMP_NUM_THREADS: 16I_MPI_PIN_DOMAIN: auto:scatterOMP_PLACES: threadsOMP_PROC_BIND: spread
Run 2x32	OMP_NUM_THREADS: 32I_MPI_PIN_DOMAIN: auto:scatterOMP_PLACES: threadsOMP_PROC_BIND: spread
Run 2x64	OMP_NUM_THREADS: 64I_MPI_PIN_DOMAIN: auto:scatterOMP_PLACES: threadsOMP_PROC_BIND: spread
Run 2x96	OMP_NUM_THREADS: 96I_MPI_PIN_DOMAIN: auto:scatterOMP_PLACES: threadsOMP_PROC_BIND: spread

(2x1) Efficiency	(2x1) Potential Speed-Up (%)	(2x2) Efficiency	(2x2) Potential Speed-Up (%)	(2x4) Efficiency	(2x4) Potential Speed-Up (%)	(2x8) Efficiency	(2x8) Potential Speed-Up (%)	(2x16) Efficiency	(2x16) Potential Speed-Up (%)	(2x32) Efficiency	(2x32) Potential Speed-Up (%)	(2x64) Efficiency	(2x64) Potential Speed-Up (%)	(2x96) Efficiency	(2x96) Potential Speed-Up (%)
1	0	0.58	0.34	0.31	0.53	0.16	0.69	0.06	0.81	0.02	1.01	0	1.23	0	1.53

Speed-Up and Efficiency

▶Detailed Speed-Up and Efficiency

Run	Number of threads	Efficiency (ideal is 1)	Speedup	Ideal Speedup	Time (s)	Coverage (%)
2x1	2	1	1	1	0.38	0.81
2x2	4	0.58	0.97	1.6666666666667	0.39	0.81
2x4	8	0.31	1.03	3.3333333333333	0.41	0.76
2x8	16	0.16	0.92	5.6666666666667	0.44	0.82
2x16	32	0.06	0.67	11.333333333333	0.62	0.86
2x32	64	0.02	0.48	22	0.92	1.03
2x64	128	0	0.21	43.333333333333	1.97	1.24
2x96	192	0	0.12	64.666666666667	3.97	1.53

Report Configuration

Code clean check

Workaround

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

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

Code clean check

Workaround

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

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

▶Scalability Runs Description

▼Colums Filter

Speed-Up and Efficiency

▶Detailed Speed-Up and Efficiency