OV - exec - Loop 99

	Loop Id: 99	Module: exec	Source: ljForce.c:158-162 [...]	Coverage: 0.39%

0x40b820 LEA	-0x38(%R10,%R11,1),%R14

0x40b825 VMOVUPS	%XMM0,(%R9)    [1]

0x40b82a MOVQ	$0,0x10(%R9)    [1]

0x40b832 ADD	$0x40,%R11

0x40b836 MOVQ	$0,(%R14)    [2]

0x40b83d VMOVUPS	%XMM0,0x18(%R9)    [1]

0x40b843 MOVQ	$0,0x28(%R9)    [1]

0x40b84b MOVQ	$0,0x8(%R14)    [2]

0x40b853 VMOVUPS	%XMM0,0x30(%R9)    [1]

0x40b859 MOVQ	$0,0x40(%R9)    [1]

0x40b861 MOVQ	$0,0x10(%R14)    [2]

0x40b869 VMOVUPS	%XMM0,0x48(%R9)    [1]

0x40b86f MOVQ	$0,0x58(%R9)    [1]

0x40b877 MOVQ	$0,0x18(%R14)    [2]

0x40b87f VMOVUPS	%XMM0,0x60(%R9)    [1]

0x40b885 MOVQ	$0,0x70(%R9)    [1]

0x40b88d MOVQ	$0,0x20(%R14)    [2]

0x40b895 VMOVUPS	%XMM0,0x78(%R9)    [1]

0x40b89b MOVQ	$0,0x88(%R9)    [1]

0x40b8a6 MOVQ	$0,0x28(%R14)    [2]

0x40b8ae VMOVUPS	%XMM0,0x90(%R9)    [1]

0x40b8b7 MOVQ	$0,0xa0(%R9)    [1]

0x40b8c2 MOVQ	$0,0x30(%R14)    [2]

0x40b8ca VMOVUPS	%XMM0,0xa8(%R9)    [1]

0x40b8d3 MOVQ	$0,0xb8(%R9)    [1]

0x40b8de ADD	$0xc0,%R9

0x40b8e5 MOVQ	$0,0x38(%R14)    [2]

0x40b8ed CMP	%R11,%R8

0x40b8f0 JNE	40b820

/beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c: 158 - 162

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

158:    for (int ii=0; ii<fSize; ++ii)

159:    {

160:       zeroReal3(s->atoms->f[ii]);

161:       s->atoms->U[ii] = 0.;

162:    }

/beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/build/CoMD/CoMD/src-openmp/mytype.h: 23 - 23

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

23:    a[1] = 0.0;

Coverage (%)	Name	Source Location	Module
►97.92+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	ljForce	ljForce.c:157	exec
○	timestep	timestep.c:51	exec
○	main	CoMD.c:125	exec
►1.04+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	ljForce	ljForce.c:157	exec
○	main	CoMD.c:211	exec
○	__libc_start_call_main		libc.so.6
►1.00+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	ljForce	ljForce.c:157	exec
○	timestep	timestep.c:51	exec
○	main	CoMD.c:125	exec
○	__libc_start_call_main		libc.so.6

Coverage (%)	Name	Source Location	Module
►100.00+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so

Coverage (%)	Name	Source Location	Module
►100.00+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so

Coverage (%)	Name	Source Location	Module
►99.89+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so

Coverage (%)	Name	Source Location	Module
►99.98+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so

Coverage (%)	Name	Source Location	Module
►99.99+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.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	1.56
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	6.25
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.56
Bottlenecks	P6, P7, P8, P9,
Function	ljForce.extracted.27
Source	ljForce.c:158-162,mytype.h:23-23
Source loop unroll info	unrolled by 8
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	8
CQA cycles	6.25
CQA cycles if no scalar integer	4.00
CQA cycles if FP arith vectorized	6.25
CQA cycles if fully vectorized	1.00
Front-end cycles	3.63
P0 cycles	0.67
P1 cycles	0.67
P2 cycles	0.67
P3 cycles	0.67
P4 cycles	0.67
P5 cycles	0.67
P6 cycles	6.25
P7 cycles	6.25
P8 cycles	6.25
P9 cycles	6.25
P10 cycles	0.00
P11 cycles	0.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	4.00
P15 cycles	4.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	29.00
Nb uops	29.00
Nb loads	0.00
Nb stores	24.00
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	40.96
Bytes prefetched	0.00
Bytes loaded	0.00
Bytes stored	256.00
Stride 0	0.00
Stride 1	2.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	33.33
Vectorization ratio load	NA
Vectorization ratio store	33.33
Vectorization ratio mul	NA
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	NA
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	NA

Metric	Value
CQA speedup if no scalar integer	1.56
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	6.25
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.56
Bottlenecks	P6, P7, P8, P9,
Function	ljForce.extracted.27
Source	ljForce.c:158-162,mytype.h:23-23
Source loop unroll info	unrolled by 8
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	8
CQA cycles	6.25
CQA cycles if no scalar integer	4.00
CQA cycles if FP arith vectorized	6.25
CQA cycles if fully vectorized	1.00
Front-end cycles	3.63
P0 cycles	0.67
P1 cycles	0.67
P2 cycles	0.67
P3 cycles	0.67
P4 cycles	0.67
P5 cycles	0.67
P6 cycles	6.25
P7 cycles	6.25
P8 cycles	6.25
P9 cycles	6.25
P10 cycles	0.00
P11 cycles	0.00
P12 cycles	0.00
P13 cycles	0.00
P14 cycles	4.00
P15 cycles	4.00
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	29.00
Nb uops	29.00
Nb loads	0.00
Nb stores	24.00
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	40.96
Bytes prefetched	0.00
Bytes loaded	0.00
Bytes stored	256.00
Stride 0	0.00
Stride 1	2.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	33.33
Vectorization ratio load	NA
Vectorization ratio store	33.33
Vectorization ratio mul	NA
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	NA
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	NA

Path /

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

Function	ljForce.extracted.27
Source file and lines	ljForce.c:158-162
Module	exec

The loop is defined in:

/beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:158-162
/beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/build/CoMD/CoMD/src-openmp/mytype.h:23

It is main loop of related source loop which is unrolled by 8 (including vectorization).

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 6.25 to 4.00 cycles (1.56x speedup).

Workaround

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

Vectorization

Your loop is poorly vectorized. Only 12% of vector register length is used (average across all SSE/AVX instructions). By fully vectorizing your loop, you can lower the cost of an iteration from 6.25 to 1.00 cycles (6.25x speedup).

Details

33% of SSE/AVX instructions are used in vector version (process two or more data elements in vector registers):

33% of SSE/AVX stores 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:
- recompile with ffast-math (included in Ofast) to extend loop vectorization to 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)

Execution units bottlenecks

Performance is limited by 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 6.25 to 4.00 cycles (1.56x speedup).

Workaround

Write less array elements
Provide more information to your compiler:
- hardcode the bounds of the corresponding 'for' loop
- use the 'restrict' C99 keyword

No data for this section

Complex instructions

Detected COMPLEX INSTRUCTIONS.

Details

These instructions generate more than one micro-operation and only one of them can be decoded during a cycle and the extra micro-operations increase pressure on execution units.

LEA: 1 occurrences►
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:161

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 storing 256 bytes.

General properties

nb instructions	29
nb uops	29
loop length	214
used x86 registers	5
used mmx registers	0
used xmm registers	1
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	3.63 cycles
front end	3.63 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.67	0.67	0.67	0.67	0.67	0.67	6.25	6.25	6.25	6.25	0.00	0.00	0.00	0.00	4.00	4.00
cycles	0.67	0.67	0.67	0.67	0.67	0.67	6.25	6.25	6.25	6.25	0.00	0.00	0.00	0.00	4.00	4.00

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	3.63
Dispatch	6.25
Data deps.	1.00
Overall L1	6.25

Vectorization ratios

INT

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

all	100%
load	NA (no load vectorizable/vectorized instructions)
store	100%
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	NA (no other vectorizable/vectorized instructions)

INT+FP

all	33%
load	NA (no load vectorizable/vectorized instructions)
store	33%
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	NA (no other vectorizable/vectorized instructions)

Vector efficiency ratios

INT

all	6%
load	NA (no load vectorizable/vectorized instructions)
store	6%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

all	25%
load	NA (no load vectorizable/vectorized instructions)
store	25%
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	NA (no other vectorizable/vectorized instructions)

INT+FP

all	12%
load	NA (no load vectorizable/vectorized instructions)
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	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 6.25 cycles. At this rate:

32% of peak store performance is reached (40.96 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: 40b820

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P14	P15	Latency	Recip. throughput	Vectorization
LEA -0x38(%R10,%R11,1),%R14	2	0	0	0.50	0.50	0	0	0.25	0.25	0.25	0.25	0	0	2	0.50	N/A
VMOVUPS %XMM0,(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x10(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
ADD $0x40,%R11	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
MOVQ $0,(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x18(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x28(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x8(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x30(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x40(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x10(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x48(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x58(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x18(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x60(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x70(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x20(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x78(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x88(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x28(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x90(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0xa0(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x30(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0xa8(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0xb8(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
ADD $0xc0,%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
MOVQ $0,0x38(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
CMP %R11,%R8	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
JNE 40b820 <ljForce.extracted.27+0xa0>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0.33-1	N/A

Function	ljForce.extracted.27
Source file and lines	ljForce.c:158-162
Module	exec

The loop is defined in:

/beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:158-162
/beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/build/CoMD/CoMD/src-openmp/mytype.h:23

It is main loop of related source loop which is unrolled by 8 (including vectorization).

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 6.25 to 4.00 cycles (1.56x speedup).

Workaround

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

Vectorization

Details

33% of SSE/AVX instructions are used in vector version (process two or more data elements in vector registers):

33% of SSE/AVX stores 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:
- recompile with ffast-math (included in Ofast) to extend loop vectorization to 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)

Execution units bottlenecks

Workaround

Write less array elements
Provide more information to your compiler:
- hardcode the bounds of the corresponding 'for' loop
- use the 'restrict' C99 keyword

No data for this section

Complex instructions

Detected COMPLEX INSTRUCTIONS.

Details

These instructions generate more than one micro-operation and only one of them can be decoded during a cycle and the extra micro-operations increase pressure on execution units.

LEA: 1 occurrences►
- /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:161

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 storing 256 bytes.

General properties

nb instructions	29
nb uops	29
loop length	214
used x86 registers	5
used mmx registers	0
used xmm registers	1
used ymm registers	0
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	3.63 cycles
front end	3.63 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.67	0.67	0.67	0.67	0.67	0.67	6.25	6.25	6.25	6.25	0.00	0.00	0.00	0.00	4.00	4.00
cycles	0.67	0.67	0.67	0.67	0.67	0.67	6.25	6.25	6.25	6.25	0.00	0.00	0.00	0.00	4.00	4.00

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	3.63
Dispatch	6.25
Data deps.	1.00
Overall L1	6.25

Vectorization ratios

INT

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

all	100%
load	NA (no load vectorizable/vectorized instructions)
store	100%
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	NA (no other vectorizable/vectorized instructions)

INT+FP

all	33%
load	NA (no load vectorizable/vectorized instructions)
store	33%
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	NA (no other vectorizable/vectorized instructions)

Vector efficiency ratios

INT

all	6%
load	NA (no load vectorizable/vectorized instructions)
store	6%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
other	NA (no other vectorizable/vectorized instructions)

all	25%
load	NA (no load vectorizable/vectorized instructions)
store	25%
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	NA (no other vectorizable/vectorized instructions)

INT+FP

all	12%
load	NA (no load vectorizable/vectorized instructions)
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	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 6.25 cycles. At this rate:

32% of peak store performance is reached (40.96 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: 40b820

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P14	P15	Latency	Recip. throughput	Vectorization
LEA -0x38(%R10,%R11,1),%R14	2	0	0	0.50	0.50	0	0	0.25	0.25	0.25	0.25	0	0	2	0.50	N/A
VMOVUPS %XMM0,(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x10(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
ADD $0x40,%R11	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
MOVQ $0,(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x18(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x28(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x8(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x30(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x40(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x10(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x48(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x58(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x18(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x60(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x70(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x20(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x78(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0x88(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x28(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0x90(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0xa0(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
MOVQ $0,0x30(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
VMOVUPS %XMM0,0xa8(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0.50	0.50	4	0.50	vect (25.0%)
MOVQ $0,0xb8(%R9)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
ADD $0xc0,%R9	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
MOVQ $0,0x38(%R14)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	4	0.50	scal (6.3%)
CMP %R11,%R8	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	1	0.17	N/A
JNE 40b820 <ljForce.extracted.27+0xa0>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0.33-1	N/A

▶Scalability Runs Description

Run 8x1	Number processes: 8Number nodes: NANumber processes per node: 8Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> --bind-to core --map-by package:PE=24 --rank-by fill --report-bindings Dataset: Run Directory: /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/run/oneview_runs/multicore/icx_1/oneview_run_1773746383OMP_PROC_BIND: spreadOMP_DISPLAY_AFFINITY: TRUEOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEOMP_PLACES: threadsOMP_NUM_THREADS: 1
Run 8x8	Number processes: 8Number processes per node: 8Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> --bind-to core --map-by package:PE=24 --rank-by fill --report-bindings Dataset: Run Directory: /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/run/oneview_runs/multicore/icx_1/oneview_run_1773746383OMP_NUM_THREADS: 8OMP_PROC_BIND: spreadOMP_DISPLAY_AFFINITY: TRUEOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEOMP_PLACES: threads
Run 8x12	Number processes: 8Number processes per node: 8Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> --bind-to core --map-by package:PE=24 --rank-by fill --report-bindings Dataset: Run Directory: /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/run/oneview_runs/multicore/icx_1/oneview_run_1773746383OMP_NUM_THREADS: 12OMP_PROC_BIND: spreadOMP_DISPLAY_AFFINITY: TRUEOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEOMP_PLACES: threads
Run 8x16	Number processes: 8Number processes per node: 8Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> --bind-to core --map-by package:PE=24 --rank-by fill --report-bindings Dataset: Run Directory: /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/run/oneview_runs/multicore/icx_1/oneview_run_1773746383OMP_NUM_THREADS: 16OMP_PROC_BIND: spreadOMP_DISPLAY_AFFINITY: TRUEOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEOMP_PLACES: threads
Run 8x20	Number processes: 8Number processes per node: 8Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> --bind-to core --map-by package:PE=24 --rank-by fill --report-bindings Dataset: Run Directory: /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/run/oneview_runs/multicore/icx_1/oneview_run_1773746383OMP_NUM_THREADS: 20OMP_PROC_BIND: spreadOMP_DISPLAY_AFFINITY: TRUEOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEOMP_PLACES: threads
Run 8x24	Number processes: 8Number processes per node: 8Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> --bind-to core --map-by package:PE=24 --rank-by fill --report-bindings Dataset: Run Directory: /beegfs/hackathon/users/eoseret/qaas_runs_test/gmz12.benchmarkcenter.megware.com/177-374-1600/CoMD/run/oneview_runs/multicore/icx_1/oneview_run_1773746383OMP_NUM_THREADS: 24OMP_PROC_BIND: spreadOMP_DISPLAY_AFFINITY: TRUEOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEOMP_PLACES: threads

(8x1) Efficiency	(8x1) Potential Speed-Up (%)	(8x8) Efficiency	(8x8) Potential Speed-Up (%)	(8x12) Efficiency	(8x12) Potential Speed-Up (%)	(8x16) Efficiency	(8x16) Potential Speed-Up (%)	(8x20) Efficiency	(8x20) Potential Speed-Up (%)	(8x24) Efficiency	(8x24) Potential Speed-Up (%)
1	0	0.35	1	0.2	1.64	0.14	2.01	0.11	2.28	0.09	2.56

Speed-Up and Efficiency

▶Detailed Speed-Up and Efficiency

Run	Number of threads	Efficiency (ideal is 1)	Speedup	Ideal Speedup	Time (s)	Coverage (%)
8x1	8	1	1	1	1.5750000476837	0.38993239402771
8x8	40	0.35	1.77	5	0.91000002622604	1.5519783496857
8x12	72	0.2	1.78	9	0.91000002622604	2.0484166145325
8x16	104	0.14	1.87	13	0.92500007152557	2.3425784111023
8x20	136	0.11	1.84	17	0.89499998092651	2.5564820766449
8x24	168	0.09	1.83	21	0.9099999666214	2.8030905723572

Report Configuration

Code clean check

Workaround

Vectorization

Details

Workaround

Execution units bottlenecks

Workaround

Complex instructions

Details

Type of elements and instruction set

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

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

Complex instructions

Details

Type of elements and instruction set

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

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

▼Columns Filter

Speed-Up and Efficiency

▶Detailed Speed-Up and Efficiency