Simulation-Based Performance Prediction of HPC Applications: A Case Study of HPL

Xu, Gen; Ibeid, Huda; Jiang, Xin; Svilan, Vjekoslav; Bian, Zhaojuan

doi:10.1109/hustprotools51951.2020.00016

Cited by 6 publications

(2 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Optimization based on performance estimation. There are many previous studies on the performance estimation of HPL [5,7,8,16,37,38]. We further adapt the performance model to estimate the execution time of SnuHPL accurately.…”

Section: Load Balancing Endo Et Al Propose An Optimization Techniquementioning

confidence: 99%

SnuHPL

Kim

Kwon

Kang

et al. 2022

Proceedings of the 36th ACM International Conference on Supercomputing

View full text Add to dashboard Cite

These days, it is typical for a large-scale cluster system to have different kinds of GPUs. However, HPL (High-Performance LINPACK), the de-facto standard LINPACK implementation for evaluating the performance of a cluster system, is originally designed to work only for homogeneous CPU-only systems. In this paper, we develop SnuHPL, an optimized HPL for clusters of modern heterogeneous GPUs. To optimize SnuHPL for the heterogeneous GPUs, we design a performance model, a SnuHPL simulator based on the model, and a greedy heuristic algorithm based on the simulator. The algorithm generates the best data distribution for a given cluster configuration by considering computing power, memory capacity, and network performance altogether. We also present a simple technique to increase the energy efficiency of HPL by adjusting the core clock frequency of the GPUs. The evaluation of the data distribution algorithm on small clusters of different GPU combinations shows that it outperforms well-known other data distribution strategies. We show the effectiveness of SnuHPL on a cluster of 1,760 NVIDIA A100-80GB GPUs and 440 A100-40GB GPUs. We also show the effectiveness of the proposed energy optimization technique on a cluster of 144 A100-80GB GPUs.CCS Concepts: • Computing methodologies → Massively parallel algorithms.

show abstract

Section: Load Balancing Endo Et Al Propose An Optimization Techniquementioning

confidence: 99%

SnuHPL

Kim

Kwon

Kang

et al. 2022

Proceedings of the 36th ACM International Conference on Supercomputing

View full text Add to dashboard Cite

show abstract

“…Skeletons are code extractions of the most important parts of a com-plex application whereas we only modify a few dozens of lines of HPL before emulating it with SMPI. Some researchers from Intel unaware of our recent work recently applied the same methodology as the one we proposed in [5] to both Intel HPL and OpenHPL in the closed-source CoFluent simulator [23].…”

Section: Related Workmentioning

confidence: 99%

Simulation-based Optimization and Sensibility Analysis of MPI Applications: Variability Matters

Cornebize¹,

Legrand²

2021

Preprint

View full text Add to dashboard Cite

Finely tuning MPI applications and understanding the influence of key parameters (number of processes, granularity, collective operation algorithms, virtual topology, and process placement) is critical to obtain good performance on supercomputers. With the high consumption of running applications at scale, doing so solely to optimize their performance is particularly costly. Having inexpensive but faithful predictions of expected performance could be a great help for researchers and system administrators. The methodology we propose decouples the complexity of the platform, which is captured through statistical models of the performance of its main components (MPI communications, BLAS operations), from the complexity of adaptive applications by emulating the application and skipping regular non-MPI parts of the code. We demonstrate the capability of our method with High-Performance Linpack (HPL), the benchmark used to rank supercomputers in the TOP500, which requires careful tuning. We briefly present ( 1) how the open-source version of HPL can be slightly modified to allow a fast emulation on a single commodity server at the scale of a supercomputer. Then we present (2) an extensive (in)validation study that compares simulation with real experiments and demonstrates our ability to predict the performance of HPL within a few percent consistently. This study allows us to identify the main modeling pitfalls (e.g., spatial and temporal node variability or network heterogeneity and irregular behavior) that need to be considered. Last, we show (3) how our "surrogate" allows studying several subtle HPL parameter optimization problems while accounting for uncertainty on the platform.

show abstract

Building a Fine-Grained Analytical Performance Model for Complex Scientific Simulations

Dijk¹,

Závodszky²,

Varbanescu³

et al. 2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Analytical performance models are powerful for understanding and predicting the performance of large-scale simulations. As such, they can help identify performance bottlenecks, assess the effect of load imbalance, or indicate performance behavior expectations when migrating to larger systems. Existing automated methods either focus on broad metrics and/or problems - e.g., application scalability behavior on large scale systems and inputs - or use black-box models that are more difficult to interpret e.g., machine-learning models.In this work we propose a methodology for building per-process analytical performance models relying on code analysis to derive a simple, high-level symbolic application model, and using empirical data to further calibrate and validate the model for accurate predictions.We demonstrate our model-building methodology on HemoCell, a high-performance framework for cell-based bloodflow simulations. We calibrate the model for two large-scale systems, with different architectures. Our results show good prediction accuracy for four different scenarios, including load-balanced configurations (average error of 3.6%, and a maximum error below 13%), and load-imbalanced ones (with an average prediction error of 10% and a maximum error below 16%).

show abstract

Simulation-Based Performance Prediction of HPC Applications: A Case Study of HPL

Cited by 6 publications

References 17 publications

SnuHPL

SnuHPL

Simulation-based Optimization and Sensibility Analysis of MPI Applications: Variability Matters

Building a Fine-Grained Analytical Performance Model for Complex Scientific Simulations

Contact Info

Product

Resources

About