OV - exec - Loop 104

	Loop Id: 104	Module: exec	Source: timestep.c:88-94	Coverage: 0.33%

0x410010 VPMOVSXDQ	(%R13,%RBX,4),%ZMM26    [4]

0x410018 VXORPD	%XMM27,%XMM27,%XMM27

0x41001e KXNORW	%K0,%K0,%K1

0x410022 VPSLLQ	$0x4,%ZMM26,%ZMM26

0x410029 VGATHERQPD	0x8(%R12,%ZMM26,1),%ZMM27{%K1}    [3]

0x410031 VDIVPD	%ZMM27,%ZMM1,%ZMM26

0x410037 VMOVUPD	(%R9,%RDX,1),%ZMM27    [2]

0x41003e VMOVUPD	0x40(%R9,%RDX,1),%ZMM28    [2]

0x410046 VMOVUPD	0x80(%R9,%RDX,1),%ZMM29    [2]

0x41004e VMOVAPD	%ZMM27,%ZMM30

0x410054 VPERMT2PD	%ZMM28,%ZMM4,%ZMM30

0x41005a VMOVAPD	%ZMM27,%ZMM31

0x410060 VPERMT2PD	%ZMM28,%ZMM6,%ZMM31

0x410066 VPERMT2PD	%ZMM27,%ZMM8,%ZMM28

0x41006c VMOVUPD	(%R8,%RDX,1),%ZMM17    [1]

0x410073 VMOVUPD	0x40(%R8,%RDX,1),%ZMM3    [1]

0x41007b VPERMT2PD	%ZMM29,%ZMM5,%ZMM30

0x410081 VMOVUPD	0x80(%R8,%RDX,1),%ZMM2    [1]

0x410089 VMOVAPD	%ZMM17,%ZMM27

0x41008f VPERMT2PD	%ZMM3,%ZMM4,%ZMM27

0x410095 VPERMT2PD	%ZMM29,%ZMM7,%ZMM31

0x41009b VPERMT2PD	%ZMM2,%ZMM5,%ZMM27

0x4100a1 VMOVAPD	%ZMM17,%ZMM16

0x4100a7 VPERMT2PD	%ZMM3,%ZMM6,%ZMM16

0x4100ad VPERMT2PD	%ZMM29,%ZMM9,%ZMM28

0x4100b3 VPERMT2PD	%ZMM2,%ZMM7,%ZMM16

0x4100b9 VPERMT2PD	%ZMM17,%ZMM8,%ZMM3

0x4100bf VFMADD231PD	%ZMM30,%ZMM26,%ZMM27

0x4100c5 VPERMT2PD	%ZMM2,%ZMM9,%ZMM3

0x4100cb VFMADD231PD	%ZMM31,%ZMM26,%ZMM16

0x4100d1 VFMADD231PD	%ZMM28,%ZMM26,%ZMM3

0x4100d7 VMOVAPD	%ZMM16,%ZMM2

0x4100dd VPERMT2PD	%ZMM27,%ZMM10,%ZMM2

0x4100e3 VMOVAPD	%ZMM27,%ZMM17

0x4100e9 VPERMT2PD	%ZMM16,%ZMM11,%ZMM17

0x4100ef VPERMT2PD	%ZMM16,%ZMM12,%ZMM27

0x4100f5 VPERMT2PD	%ZMM3,%ZMM13,%ZMM27

0x4100fb VPERMT2PD	%ZMM3,%ZMM14,%ZMM17

0x410101 VPERMT2PD	%ZMM3,%ZMM15,%ZMM2

0x410107 VMOVUPD	%ZMM2,0x80(%R8,%RDX,1)    [1]

0x41010f VMOVUPD	%ZMM17,0x40(%R8,%RDX,1)    [1]

0x410117 VMOVUPD	%ZMM27,(%R8,%RDX,1)    [1]

0x41011e ADD	$0x8,%RBX

0x410122 ADD	$0xc0,%RDX

0x410129 CMP	%ECX,%EBX

0x41012b JLE	410010

/home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c: 88 - 94

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

88:       for (int iOff=MAXATOMS*iBox,ii=0; ii<s->boxes->nAtoms[iBox]; ii++,iOff++)

89:       {

90:          int iSpecies = s->atoms->iSpecies[iOff];

91:          real_t invMass = 1.0/s->species[iSpecies].mass;

92:          s->atoms->r[iOff][0] += dt*s->atoms->p[iOff][0]*invMass;

93:          s->atoms->r[iOff][1] += dt*s->atoms->p[iOff][1]*invMass;

94:          s->atoms->r[iOff][2] += dt*s->atoms->p[iOff][2]*invMass;

Coverage (%)	Name	Source Location	Module
►87.35+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	timestep	timestep.c:43	exec
○	main	CoMD.c:125	exec
○	__libc_start_call_main		libc.so.6
►12.54+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	timestep	timestep.c:43	exec
○	main	CoMD.c:125	exec

Coverage (%)	Name	Source Location	Module
►90.50+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_launch_thread		libiomp5.so
○	_INTERNALea951ae6::__kmp_launc[...]		libiomp5.so
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►7.54+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	timestep	timestep.c:43	exec
○	main	CoMD.c:125	exec
►1.34+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_launch_thread		libiomp5.so
○	_INTERNALea951ae6::__kmp_launc[...]		libiomp5.so

Coverage (%)	Name	Source Location	Module
►92.41+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_launch_thread		libiomp5.so
○	_INTERNALea951ae6::__kmp_launc[...]		libiomp5.so
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►6.66+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	timestep	timestep.c:43	exec
○	main	CoMD.c:125	exec

Coverage (%)	Name	Source Location	Module
►94.22+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_launch_thread		libiomp5.so
○	_INTERNALea951ae6::__kmp_launc[...]		libiomp5.so
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►4.82+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	timestep	timestep.c:43	exec
○	main	CoMD.c:125	exec

Coverage (%)	Name	Source Location	Module
►94.97+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_launch_thread		libiomp5.so
○	_INTERNALea951ae6::__kmp_launc[...]		libiomp5.so
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►3.82+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	timestep	timestep.c:43	exec
○	main	CoMD.c:125	exec

Coverage (%)	Name	Source Location	Module
►95.33+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_launch_thread		libiomp5.so
○	_INTERNALea951ae6::__kmp_launc[...]		libiomp5.so
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►3.12+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	timestep	timestep.c:43	exec
○	main	CoMD.c:125	exec
►1.22+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_launch_thread		libiomp5.so
○	_INTERNALea951ae6::__kmp_launc[...]		libiomp5.so

Coverage (%)	Name	Source Location	Module
►96.22+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_launch_thread		libiomp5.so
○	_INTERNALea951ae6::__kmp_launc[...]		libiomp5.so
○	start_thread		libc.so.6
○	__GI___clone3		libc.so.6
►2.89+	__kmp_invoke_microtask		libiomp5.so
○	__kmp_invoke_task_func		libiomp5.so
○	__kmp_fork_call		libiomp5.so
○	__kmpc_fork_call		libiomp5.so
○	timestep	timestep.c:43	exec
○	main	CoMD.c:125	exec

min	med	avg	max

Percentile Index	10	20	30	40	50	60	70	80	90	100
Value

min	med	avg	max

Percentile Index	10	20	30	40	50	60	70	80	90	100
Value

Path /

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	1.01
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.05
Bottlenecks	micro-operation queue, P11,
Function	advancePosition.extracted
Source	timestep.c:88-94
Source loop unroll info	unrolled by 8
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	8
CQA cycles	11.50
CQA cycles if no scalar integer	11.50
CQA cycles if FP arith vectorized	11.50
CQA cycles if fully vectorized	11.38
Front-end cycles	11.50
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	0.50
P3 cycles	0.50
P4 cycles	0.50
P5 cycles	0.50
P6 cycles	4.50
P7 cycles	4.50
P8 cycles	4.50
P9 cycles	4.50
P10 cycles	8.25
P11 cycles	11.50
P12 cycles	11.00
P13 cycles	8.25
P14 cycles	6.50
P15 cycles	6.50
DIV/SQRT cycles	4.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	46.00
Nb uops	92.00
Nb loads	8.00
Nb stores	3.00
Nb stack references	0.00
FLOP/cycle	4.87
Nb FLOP add-sub	0.00
Nb FLOP mul	0.00
Nb FLOP fma	24.00
Nb FLOP div	8.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	58.43
Bytes prefetched	0.00
Bytes loaded	480.00
Bytes stored	192.00
Stride 0	0.00
Stride 1	3.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	1.00
Vectorization ratio all	100.00
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	100.00
Vectorization ratio other	100.00
Vector-efficiency ratio all	96.95
Vector-efficiency ratio load	93.75
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	100.00
Vector-efficiency ratio div_sqrt	100.00
Vector-efficiency ratio other	97.22

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	1.01
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	1.05
Bottlenecks	micro-operation queue, P11,
Function	advancePosition.extracted
Source	timestep.c:88-94
Source loop unroll info	unrolled by 8
Source loop unroll confidence level	max
Unroll/vectorization loop type	main
Unroll factor	8
CQA cycles	11.50
CQA cycles if no scalar integer	11.50
CQA cycles if FP arith vectorized	11.50
CQA cycles if fully vectorized	11.38
Front-end cycles	11.50
P0 cycles	0.50
P1 cycles	0.50
P2 cycles	0.50
P3 cycles	0.50
P4 cycles	0.50
P5 cycles	0.50
P6 cycles	4.50
P7 cycles	4.50
P8 cycles	4.50
P9 cycles	4.50
P10 cycles	8.25
P11 cycles	11.50
P12 cycles	11.00
P13 cycles	8.25
P14 cycles	6.50
P15 cycles	6.50
DIV/SQRT cycles	4.00
Inter-iter dependencies cycles	1
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	46.00
Nb uops	92.00
Nb loads	8.00
Nb stores	3.00
Nb stack references	0.00
FLOP/cycle	4.87
Nb FLOP add-sub	0.00
Nb FLOP mul	0.00
Nb FLOP fma	24.00
Nb FLOP div	8.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	58.43
Bytes prefetched	0.00
Bytes loaded	480.00
Bytes stored	192.00
Stride 0	0.00
Stride 1	3.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	1.00
Vectorization ratio all	100.00
Vectorization ratio load	100.00
Vectorization ratio store	100.00
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	100.00
Vectorization ratio div_sqrt	100.00
Vectorization ratio other	100.00
Vector-efficiency ratio all	96.95
Vector-efficiency ratio load	93.75
Vector-efficiency ratio store	100.00
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	100.00
Vector-efficiency ratio div_sqrt	100.00
Vector-efficiency ratio other	97.22

Path /

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

Function	advancePosition.extracted
Source file and lines	timestep.c:88-94
Module	exec

The loop is defined in /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:88-94.

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

gain
potential
hint
expert

Vectorization

Your loop is vectorized, but using 96% register length (average across all SSE/AVX instructions). By fully vectorizing your loop, you can lower the cost of an iteration from 11.50 to 11.37 cycles (1.01x speedup).

Details

All SSE/AVX instructions are used in vector version (process two or more data elements in vector registers). Since your execution units are vector units, only a fully vectorized loop can use their full power.

Workaround

Pass to your compiler a micro-architecture specialization option:
- use march=native
Use vector aligned instructions:
1. align your arrays on 64 bytes boundaries: replace { void *p = malloc (size); } with { void *p; posix_memalign (&p, 64, size); }.
2. inform your compiler that your arrays are vector aligned: if array 'foo' is 64 bytes-aligned, define a pointer 'p_foo' as __builtin_assume_aligned (foo, 64) and use it instead of 'foo' in the loop.

Execution units bottlenecks

Performance is limited by execution of FP multiply or FMA (fused multiply-add) operations (the FP multiply/FMA unit is a bottleneck).

Workaround

Reduce the number of FP multiply/FMA instructions

FMA

Detected 24 FMA (fused multiply-add) operations.

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.

VGATHERQPD: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:91

Slow data structures access

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

Details

Irregular (variable stride) or indirect: 1 occurrence(s)

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

Workaround

Try to remove indirect accesses. If applicable, precompute elements out of the innermost loop.

Vector unaligned load/store instructions

Detected 9 optimal vector unaligned load/store instructions.

Details

VMOVUPD: 9 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:88
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:88
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:88

Workaround

Use vector aligned instructions:

align your arrays on 64 bytes boundaries: replace { void *p = malloc (size); } with { void *p; posix_memalign (&p, 64, size); }.
inform your compiler that your arrays are vector aligned: if array 'foo' is 64 bytes-aligned, define a pointer 'p_foo' as __builtin_assume_aligned (foo, 64) and use it instead of 'foo' in the loop.

Gather/scatter instructions

Detected gather/scatter instructions (typically caused by indirect accesses).

Details

VGATHERQPD: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:91

Workaround

Try to simplify your code and/or replace indirect accesses with unit-stride ones.

Type of elements and instruction set

4 AVX-512 instructions are processing arithmetic or math operations on double precision FP elements in vector mode (eight at a time).

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

The binary loop is composed of 56 FP arithmetical operations:

24: addition or subtraction (all inside FMA instructions)
24: multiply (all inside FMA instructions)
8: divide

The binary loop is loading 480 bytes (60 double precision FP elements). The binary loop is storing 192 bytes (24 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.08 FP operations per loaded or stored byte.

General properties

nb instructions	46
nb uops	92
loop length	289
used x86 registers	7
used mmx registers	0
used xmm registers	1
used ymm registers	0
used zmm registers	23
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	11.50 cycles
front end	11.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.50	0.50	0.50	0.50	0.50	0.50	4.50	4.50	4.50	4.50	8.25	11.50	11.00	8.25	6.50	6.50
cycles	0.50	0.50	0.50	0.50	0.50	0.50	4.50	4.50	4.50	4.50	8.25	11.50	11.00	8.25	6.50	6.50

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	4.00
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	11.50
Dispatch	11.50
DIV/SQRT	4.00
Data deps.	1.00
Overall L1	11.50

Vectorization ratios

INT

all	100%
load	100%
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)
other	100%

all	100%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	100%
div/sqrt	100%
other	100%

INT+FP

all	100%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	100%
div/sqrt	100%
other	100%

Vector efficiency ratios

INT

all	75%
load	50%
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)
other	100%

all	98%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	100%
div/sqrt	100%
other	97%

INT+FP

all	96%
load	93%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	100%
div/sqrt	100%
other	97%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 11.50 cycles. At this rate:

16% of peak load performance is reached (41.74 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
13% of peak store performance is reached (16.70 out of 128.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Performance is limited by instruction throughput (loading/decoding program instructions to execution core) (front-end is a bottleneck).

ASM code

In the binary file, the address of the loop is: 410010

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
VPMOVSXDQ (%R13,%RBX,4),%ZMM26	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	4	0.50	vect (50.0%)
VXORPD %XMM27,%XMM27,%XMM27	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
KXNORW %K0,%K0,%K1	1	0	0	0	0	0	0	0	0	0	0	0.50	0	0	0.50	0	0	1	0.50	N/A
VPSLLQ $0x4,%ZMM26,%ZMM26	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	1	0.50	vect (100.0%)
VGATHERQPD 0x8(%R12,%ZMM26,1),%ZMM27{%K1}	48	0	0	0	0	0	0	2	2	2	2	2.50	4	4.50	3	5	5	0-16	8.94	vect (100.0%)
VDIVPD %ZMM27,%ZMM1,%ZMM26	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	0	13	4	vect (100.0%)
VMOVUPD (%R9,%RDX,1),%ZMM27	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVUPD 0x40(%R9,%RDX,1),%ZMM28	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVUPD 0x80(%R9,%RDX,1),%ZMM29	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVAPD %ZMM27,%ZMM30	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM28,%ZMM4,%ZMM30	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVAPD %ZMM27,%ZMM31	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM28,%ZMM6,%ZMM31	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM27,%ZMM8,%ZMM28	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVUPD (%R8,%RDX,1),%ZMM17	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVUPD 0x40(%R8,%RDX,1),%ZMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VPERMT2PD %ZMM29,%ZMM5,%ZMM30	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVUPD 0x80(%R8,%RDX,1),%ZMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVAPD %ZMM17,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM4,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM29,%ZMM7,%ZMM31	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM2,%ZMM5,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVAPD %ZMM17,%ZMM16	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM6,%ZMM16	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM29,%ZMM9,%ZMM28	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM2,%ZMM7,%ZMM16	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM17,%ZMM8,%ZMM3	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VFMADD231PD %ZMM30,%ZMM26,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50	vect (100.0%)
VPERMT2PD %ZMM2,%ZMM9,%ZMM3	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VFMADD231PD %ZMM31,%ZMM26,%ZMM16	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50	vect (100.0%)
VFMADD231PD %ZMM28,%ZMM26,%ZMM3	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50	vect (100.0%)
VMOVAPD %ZMM16,%ZMM2	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM27,%ZMM10,%ZMM2	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVAPD %ZMM27,%ZMM17	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM16,%ZMM11,%ZMM17	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM16,%ZMM12,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM13,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM14,%ZMM17	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM15,%ZMM2	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVUPD %ZMM2,0x80(%R8,%RDX,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	1	vect (100.0%)
VMOVUPD %ZMM17,0x40(%R8,%RDX,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	1	vect (100.0%)
VMOVUPD %ZMM27,(%R8,%RDX,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	1	vect (100.0%)
ADD $0x8,%RBX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
ADD $0xc0,%RDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %ECX,%EBX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
JLE 410010 <advancePosition.extracted+0x430>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A

Function	advancePosition.extracted
Source file and lines	timestep.c:88-94
Module	exec

gain
potential
hint
expert

Vectorization

Details

Workaround

Pass to your compiler a micro-architecture specialization option:
- use march=native
Use vector aligned instructions:
1. align your arrays on 64 bytes boundaries: replace { void *p = malloc (size); } with { void *p; posix_memalign (&p, 64, size); }.
2. inform your compiler that your arrays are vector aligned: if array 'foo' is 64 bytes-aligned, define a pointer 'p_foo' as __builtin_assume_aligned (foo, 64) and use it instead of 'foo' in the loop.

Execution units bottlenecks

Performance is limited by execution of FP multiply or FMA (fused multiply-add) operations (the FP multiply/FMA unit is a bottleneck).

Workaround

Reduce the number of FP multiply/FMA instructions

FMA

Detected 24 FMA (fused multiply-add) operations.

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.

VGATHERQPD: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:91

Slow data structures access

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

Details

Irregular (variable stride) or indirect: 1 occurrence(s)

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

Workaround

Try to remove indirect accesses. If applicable, precompute elements out of the innermost loop.

Vector unaligned load/store instructions

Detected 9 optimal vector unaligned load/store instructions.

Details

VMOVUPD: 9 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:92
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:88
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:88
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:88

Workaround

Use vector aligned instructions:

align your arrays on 64 bytes boundaries: replace { void *p = malloc (size); } with { void *p; posix_memalign (&p, 64, size); }.
inform your compiler that your arrays are vector aligned: if array 'foo' is 64 bytes-aligned, define a pointer 'p_foo' as __builtin_assume_aligned (foo, 64) and use it instead of 'foo' in the loop.

Gather/scatter instructions

Detected gather/scatter instructions (typically caused by indirect accesses).

Details

VGATHERQPD: 1 occurrences►
- /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:91

Workaround

Try to simplify your code and/or replace indirect accesses with unit-stride ones.

Type of elements and instruction set

4 AVX-512 instructions are processing arithmetic or math operations on double precision FP elements in vector mode (eight at a time).

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

The binary loop is composed of 56 FP arithmetical operations:

24: addition or subtraction (all inside FMA instructions)
24: multiply (all inside FMA instructions)
8: divide

The binary loop is loading 480 bytes (60 double precision FP elements). The binary loop is storing 192 bytes (24 double precision FP elements).

Arithmetic intensity

Arithmetic intensity is 0.08 FP operations per loaded or stored byte.

General properties

nb instructions	46
nb uops	92
loop length	289
used x86 registers	7
used mmx registers	0
used xmm registers	1
used ymm registers	0
used zmm registers	23
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	11.50 cycles
front end	11.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15
uops	0.50	0.50	0.50	0.50	0.50	0.50	4.50	4.50	4.50	4.50	8.25	11.50	11.00	8.25	6.50	6.50
cycles	0.50	0.50	0.50	0.50	0.50	0.50	4.50	4.50	4.50	4.50	8.25	11.50	11.00	8.25	6.50	6.50

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	4.00
Longest recurrence chain latency (RecMII)	1.00

Cycles summary

Front-end	11.50
Dispatch	11.50
DIV/SQRT	4.00
Data deps.	1.00
Overall L1	11.50

Vectorization ratios

INT

all	100%
load	100%
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)
other	100%

all	100%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	100%
div/sqrt	100%
other	100%

INT+FP

all	100%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	100%
div/sqrt	100%
other	100%

Vector efficiency ratios

INT

all	75%
load	50%
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)
other	100%

all	98%
load	100%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	100%
div/sqrt	100%
other	97%

INT+FP

all	96%
load	93%
store	100%
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	100%
div/sqrt	100%
other	97%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 11.50 cycles. At this rate:

16% of peak load performance is reached (41.74 out of 256.00 bytes loaded per cycle (GB/s @ 1GHz))
13% of peak store performance is reached (16.70 out of 128.00 bytes stored per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Performance is limited by instruction throughput (loading/decoding program instructions to execution core) (front-end is a bottleneck).

ASM code

In the binary file, the address of the loop is: 410010

Instruction	Nb FU	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11	P12	P13	P14	P15	Latency	Recip. throughput	Vectorization
VPMOVSXDQ (%R13,%RBX,4),%ZMM26	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0.50	0.50	0	0	0	4	0.50	vect (50.0%)
VXORPD %XMM27,%XMM27,%XMM27	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (25.0%)
KXNORW %K0,%K0,%K1	1	0	0	0	0	0	0	0	0	0	0	0.50	0	0	0.50	0	0	1	0.50	N/A
VPSLLQ $0x4,%ZMM26,%ZMM26	1	0	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	1	0.50	vect (100.0%)
VGATHERQPD 0x8(%R12,%ZMM26,1),%ZMM27{%K1}	48	0	0	0	0	0	0	2	2	2	2	2.50	4	4.50	3	5	5	0-16	8.94	vect (100.0%)
VDIVPD %ZMM27,%ZMM1,%ZMM26	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	0	13	4	vect (100.0%)
VMOVUPD (%R9,%RDX,1),%ZMM27	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVUPD 0x40(%R9,%RDX,1),%ZMM28	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVUPD 0x80(%R9,%RDX,1),%ZMM29	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVAPD %ZMM27,%ZMM30	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM28,%ZMM4,%ZMM30	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVAPD %ZMM27,%ZMM31	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM28,%ZMM6,%ZMM31	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM27,%ZMM8,%ZMM28	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVUPD (%R8,%RDX,1),%ZMM17	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVUPD 0x40(%R8,%RDX,1),%ZMM3	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VPERMT2PD %ZMM29,%ZMM5,%ZMM30	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVUPD 0x80(%R8,%RDX,1),%ZMM2	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0	0	3	0.50	vect (100.0%)
VMOVAPD %ZMM17,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM4,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM29,%ZMM7,%ZMM31	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM2,%ZMM5,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVAPD %ZMM17,%ZMM16	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM6,%ZMM16	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM29,%ZMM9,%ZMM28	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM2,%ZMM7,%ZMM16	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM17,%ZMM8,%ZMM3	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VFMADD231PD %ZMM30,%ZMM26,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50	vect (100.0%)
VPERMT2PD %ZMM2,%ZMM9,%ZMM3	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VFMADD231PD %ZMM31,%ZMM26,%ZMM16	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50	vect (100.0%)
VFMADD231PD %ZMM28,%ZMM26,%ZMM3	1	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	0	4	0.50	vect (100.0%)
VMOVAPD %ZMM16,%ZMM2	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM27,%ZMM10,%ZMM2	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVAPD %ZMM27,%ZMM17	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.17	vect (100.0%)
VPERMT2PD %ZMM16,%ZMM11,%ZMM17	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM16,%ZMM12,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM13,%ZMM27	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM14,%ZMM17	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VPERMT2PD %ZMM3,%ZMM15,%ZMM2	1	0	0	0	0	0	0	0	0	0	0	0	0.50	0.50	0	0	0	5	0.50	vect (100.0%)
VMOVUPD %ZMM2,0x80(%R8,%RDX,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	1	vect (100.0%)
VMOVUPD %ZMM17,0x40(%R8,%RDX,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	1	vect (100.0%)
VMOVUPD %ZMM27,(%R8,%RDX,1)	1	0	0	0	0	0	0	0.25	0.25	0.25	0.25	0	0	0	0	0.50	0.50	4	1	vect (100.0%)
ADD $0x8,%RBX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
ADD $0xc0,%RDX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %ECX,%EBX	1	0.17	0.17	0.17	0.17	0.17	0.17	0	0	0	0	0	0	0	0	0	0	1	0.17	N/A
JLE 410010 <advancePosition.extracted+0x430>	1	0	0	0	0.33	0.33	0.33	0	0	0	0	0	0	0	0	0	0	1	0.50-1	N/A

▶Scalability Runs Description

Run 8x1	Number processes: 8Number nodes: 1Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> Dataset: Run Directory: /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/run/oneview_runs/multicore/icx/oneview_run_1740438801OMP_NUM_THREADS: 1I_MPI_PIN_ORDER: bunchOMP_DISPLAY_AFFINITY: TRUEOMP_PROC_BIND: spreadOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEI_MPI_PIN_DOMAIN: autoI_MPI_DEBUG: 4OMP_PLACES: threads
Run 8x12	Number processes: 8Number nodes: 1Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> Dataset: Run Directory: /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/run/oneview_runs/multicore/icx/oneview_run_1740438801OMP_NUM_THREADS: 12I_MPI_PIN_ORDER: bunchOMP_DISPLAY_AFFINITY: TRUEOMP_PROC_BIND: spreadOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEI_MPI_PIN_DOMAIN: autoI_MPI_DEBUG: 4OMP_PLACES: threads
Run 8x16	Number processes: 8Number nodes: 1Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> Dataset: Run Directory: /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/run/oneview_runs/multicore/icx/oneview_run_1740438801OMP_NUM_THREADS: 16I_MPI_PIN_ORDER: bunchOMP_DISPLAY_AFFINITY: TRUEOMP_PROC_BIND: spreadOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEI_MPI_PIN_DOMAIN: autoI_MPI_DEBUG: 4OMP_PLACES: threads
Run 8x20	Number processes: 8Number nodes: 1Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> Dataset: Run Directory: /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/run/oneview_runs/multicore/icx/oneview_run_1740438801OMP_NUM_THREADS: 20I_MPI_PIN_ORDER: bunchOMP_DISPLAY_AFFINITY: TRUEOMP_PROC_BIND: spreadOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEI_MPI_PIN_DOMAIN: autoI_MPI_DEBUG: 4OMP_PLACES: threads
Run 8x24	Number processes: 8Number nodes: 1Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> Dataset: Run Directory: /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/run/oneview_runs/multicore/icx/oneview_run_1740438801OMP_NUM_THREADS: 24I_MPI_PIN_ORDER: bunchOMP_DISPLAY_AFFINITY: TRUEOMP_PROC_BIND: spreadOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEI_MPI_PIN_DOMAIN: autoI_MPI_DEBUG: 4OMP_PLACES: threads
Run 8x28	Number processes: 8Number nodes: 1Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> Dataset: Run Directory: /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/run/oneview_runs/multicore/icx/oneview_run_1740438801OMP_NUM_THREADS: 28I_MPI_PIN_ORDER: bunchOMP_DISPLAY_AFFINITY: TRUEOMP_PROC_BIND: spreadOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEI_MPI_PIN_DOMAIN: autoI_MPI_DEBUG: 4OMP_PLACES: threads
Run 8x32	Number processes: 8Number nodes: 1Run Command: <executable> -x 200 -y 200 -z 200 --xproc=2 --yproc=2 --zproc=2MPI Command: mpirun -n <number_processes> Dataset: Run Directory: /home/eoseret/qaas_runs_ZEN5/174-043-3878/intel/CoMD/run/oneview_runs/multicore/icx/oneview_run_1740438801OMP_NUM_THREADS: 32I_MPI_PIN_ORDER: bunchOMP_DISPLAY_AFFINITY: TRUEOMP_PROC_BIND: spreadOMP_AFFINITY_FORMAT: 'OMP: pid %P tid %i thread %n bound to OS proc set {%A}'OMP_DISPLAY_ENV: TRUEI_MPI_PIN_DOMAIN: autoI_MPI_DEBUG: 4OMP_PLACES: threads

(8x1) Efficiency	(8x1) 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 (%)	(8x28) Efficiency	(8x28) Potential Speed-Up (%)	(8x32) Efficiency	(8x32) Potential Speed-Up (%)
1	0	0.68	0.13	0.53	0.23	0.47	0.27	0.4	0.33	0.35	0.39	0.28	0.48

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	1.7450000047684	0.334148645401
8x12	96	0.68	8.22	12	0.2950000166893	0.40913316607475
8x16	128	0.53	8.54	16	0.31000000238419	0.50331127643585
8x20	160	0.47	9.31	20	0.29000002145767	0.51217859983444
8x24	192	0.4	9.65	24	0.26499998569489	0.55913269519806
8x28	224	0.35	9.78	28	0.25499999523163	0.59292501211166
8x32	256	0.28	9.07	32	0.29000002145767	0.67658931016922

Report Configuration

Vectorization

Details

Workaround

Execution units bottlenecks

Workaround

FMA

Complex instructions

Details

Slow data structures access

Details

Workaround

Vector unaligned load/store instructions

Details

Workaround

Gather/scatter instructions

Details

Workaround

Type of elements and instruction set

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

Arithmetic intensity

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code

Vectorization

Details

Workaround

Execution units bottlenecks

Workaround

FMA

Complex instructions

Details

Slow data structures access

Details

Workaround

Vector unaligned load/store instructions

Details

Workaround

Gather/scatter instructions

Details

Workaround

Type of elements and instruction set

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

Arithmetic intensity

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