OV - Compare Index

Help is available by moving the cursor above any symbol or by checking MAQAO website.

▶Compared Reports

r0: OMP1 - option thread_filter-threshold (1%) discards 7 threads, cumulating 0.12 seconds CPU time.
r1: OMP2 - option thread_filter-threshold (1%) discards 8 threads, cumulating 0.11 seconds CPU time.
r2: OMP4 - option thread_filter-threshold (1%) discards 8 threads, cumulating 0.13 seconds CPU time.
r3: OMP8 - option thread_filter-threshold (1%) discards 8 threads, cumulating 0.12 seconds CPU time.
r4: OMP16 - option thread_filter-threshold (1%) discards 8 threads, cumulating 0.13 seconds CPU time.
r5: OMP24 - option thread_filter-threshold (1%) discards 8 threads, cumulating 0.12 seconds CPU time.

Global Metrics

Metric		r0	r1	r2	r3	r4	r5
Total Time (s)		748.15	423.00	249.76	164.57	113.53	91.74
Max (Thread Active Time) (s)		727.42	410.15	241.97	158.17	108.24	86.43
Average Active Time (s)		726.56	397.74	224.29	140.94	87.72	67.75
Activity Ratio (%)		97.1	94.2	90.1	86.1	78.0	74.7
Average number of active threads		7.769	15.044	28.737	54.812	98.900	141.787
Affinity Stability (%)		15.1	14.0	22.5	33.1	47.2	42.5
GFLOPS		82.850	146.385	247.933	375.890	545.237	674.333
Time in analyzed loops (%)		2.05	1.36	0.91	0.58	0.43	0.35
Time in analyzed innermost loops (%)		2.04	1.35	0.90	0.58	0.43	0.35
Time in user code (%)		99.4	95.9	90.8	86.1	73.2	68.4
Compilation Options Score (%)		100	100	100	100	100	100
Array Access Efficiency (%)		55.3	52.9	52.6	50.5	50.4	50.1

Potential Speedups
Perfect Flow Complexity		1.00	1.00	1.00	1.00	1.00	1.00
Perfect OpenMP + MPI + Pthread		1.00	1.00	1.00	1.00	1.01	1.01
Perfect OpenMP + MPI + Pthread + Perfect Load Distribution		1.00	1.07	1.19	1.30	1.68	1.86
Scalability - Gap		1.00	1.13	1.34	1.76	2.43	2.94
No Scalar Integer	Potential Speedup	1.00	1.00	1.00	1.00	1.00	1.00
No Scalar Integer	Nb Loops to get 80%	1	1	1	1	1	1
FP Vectorised	Potential Speedup	1.00	1.00	1.00	1.00	1.00	1.00
FP Vectorised	Nb Loops to get 80%	1	1	1	1	1	1
Fully Vectorised	Potential Speedup	1.01	1.01	1.01	1.00	1.00	1.00
Fully Vectorised	Nb Loops to get 80%	3	3	3	3	3	3
Only FP Arithmetic	Potential Speedup	1.02	1.01	1.01	1.00	1.00	1.00
Only FP Arithmetic	Nb Loops to get 80%	3	3	3	3	3	3

Scalability Speedup

Cumulated Speedup If No Scalar Integer

Cumulated Speedup If FP Vectorized

Cumulated Speedup If Fully Vectorized

Cumulated Speedup If Only FP Arithmetic

Loop Based Profiles

Innermost / Single Loops

Inbetween Loops

Outermost Loops

Cumulated Coverage With All Loops

Innermost Loop Based Profiles

Coverage

Count

Application Categorization

Time

Coverage

Compilation Options

Source Object	Issue
▼xhpl–
▼HPL_lmul.c–
○
▼HPL_rand.c–
○
▼HPL_dlaswp03N.c–
○
▼HPL_bcast.c–
○
▼HPL_dlacpy.c–
○
▼HPL_dlaswp04N.c–
○
▼HPL_1ring.c–
○
▼HPL_setran.c–
○
▼HPL_ladd.c–
○
▼HPL_pdlange.c–
○
▼HPL_pdgesv0.c–
○
▼HPL_dlaswp02N.c–
○
▼HPL_dlaswp01N.c–
○

Path Count Profiles

Coverage

Count

Low Iteration Count Profiles

Coverage

Count

Average Number of Active Threads

Run 1 - OMP1

Run 2 - OMP2

Run 3 - OMP4

Run 4 - OMP8

Run 5 - OMP16

Run 6 - OMP24

Experiment Summaries

	r0	r1	r2	r3	r4	r5
Experiment Name
Application	./hpl-2.3/bin/Linux_Intel64_Zen5_AOCL/xhpl	same as r0	same as r0	same as r0	same as r0	same as r0
Timestamp	2025-06-23 10:53:13	same as r0	same as r0	same as r0	same as r0	same as r0
Experiment Type	MPI;	MPI; OpenMP;	same as r1	same as r1	same as r1	same as r1
Machine	gmz12.benchmarkcenter.megware.com	same as r0	same as r0	same as r0	same as r0	same as r0
Architecture	x86_64	same as r0	same as r0	same as r0	same as r0	same as r0
Micro Architecture	ZEN_V5	same as r0	same as r0	same as r0	same as r0	same as r0
Model Name	AMD EPYC 9655 96-Core Processor	same as r0	same as r0	same as r0	same as r0	same as r0
Cache Size	1024 KB	same as r0	same as r0	same as r0	same as r0	same as r0
Number of Cores	96	same as r0	same as r0	same as r0	same as r0	same as r0
Maximal Frequency	4.509375 GHz	same as r0	same as r0	same as r0	same as r0	same as r0
OS Version	Linux 5.14.0-503.31.1.el9_5.x86_64 #1 SMP PREEMPT_DYNAMIC Thu Mar 13 06:50:51 EDT 2025	same as r0	same as r0	same as r0	same as r0	same as r0
Architecture used during static analysis	x86_64	same as r0	same as r0	same as r0	same as r0	same as r0
Micro Architecture used during static analysis	ZEN_V5	same as r0	same as r0	same as r0	same as r0	same as r0
Compilation Options	xhpl: AMD clang version 17.0.6 (CLANG: AOCC_5.0.0-Build#1377 2024_09_24) /cluster/comp/aocc/5.0.0/bin/clang-17 -o HPL_rand.o -c -D Add__ -D F77_INTEGER=int -D StringSunStyle -D HPL_DETAILED_TIMING -D HPL_PROGRESS_REPORT -I /beegfs/hackathon/users/eoseret/linpack/hpl-2.3/include -I /beegfs/hackathon/users/eoseret/linpack/hpl-2.3/include/Linux_Intel64_Zen5_AOCL -I /cluster/libs/aocl/5.0.0/aocc/include -fopenmp -O3 -ffast-math -g -grecord-command-line -march=znver5 -mprefer-vector-width=512 -Wall ../HPL_rand.c -I /cluster/hpcx/2.22/ompi-aocc/include -I /cluster/hpcx/2.22/ompi-aocc/include/openmpi -I /cluster/hpcx/2.22/ompi-aocc/include/openmpi/opal/mca/event/libevent2022/libevent -I /cluster/hpcx/2.22/ompi-aocc/include/openmpi/opal/mca/event/libevent2022/libevent/include	same as r0	same as r0	same as r0	same as r0	same as r0
Number of processes observed	8	same as r0	same as r0	same as r0	same as r0	same as r0
Number of threads observed	8	16	32	64	128	192
Frequency Driver	acpi-cpufreq	same as r0	same as r0	same as r0	same as r0	same as r0
Frequency Governor	ondemand	same as r0	same as r0	same as r0	same as r0	same as r0
Huge Pages	always	same as r0	same as r0	same as r0	same as r0	same as r0
Hyperthreading	on	same as r0	same as r0	same as r0	same as r0	same as r0
Number of sockets	2	same as r0	same as r0	same as r0	same as r0	same as r0
Number of cores per socket	96	same as r0	same as r0	same as r0	same as r0	same as r0
MAQAO version	2025.1.0	same as r0	same as r0	same as r0	same as r0	same as r0
MAQAO build	b107544c0173fc3785aa7d997ff783dc12b975d2::20250527-133805	same as r0	same as r0	same as r0	same as r0	same as r0
Comments	HPL benchmark compiled with AMD AOCC/AOCL 5.0. Matrix order: 100K, block size 384. Run on AMD Zen 5 with 8 NUMA nodes and 24 cores per NUMA node	same as r0	same as r0	same as r0	same as r0	same as r0

Report Configuration

▶Compared Reports

Global Metrics

Scalability Speedup

Cumulated Speedup If No Scalar Integer

Cumulated Speedup If FP Vectorized

Cumulated Speedup If Fully Vectorized

Cumulated Speedup If Only FP Arithmetic

Loop Based Profiles

Innermost / Single Loops

Inbetween Loops

Outermost Loops

Cumulated Coverage With All Loops

Innermost Loop Based Profiles

Coverage

Count

Application Categorization

Time

Coverage

Compilation Options

Path Count Profiles

Coverage

Count

Low Iteration Count Profiles

Coverage

Count

Average Number of Active Threads

Run 1 - OMP1

Run 2 - OMP2

Run 3 - OMP4

Run 4 - OMP8

Run 5 - OMP16

Run 6 - OMP24

Experiment Summaries