Proceedings of the 2022 ACM/SPEC

It is our great pleasure to welcome you to the 13th annual ACM/SPEC International Conference on Performance Engineering (ICPE)!

Planning for ICPE'22 has started early in 2021, with a landmark decision by the steering committee to hold the conference in China, for the first time in the conference's history. Due to the ongoing COVID-19 pandemic, a decision was made to strive for a hybrid conference, with local sessions in Beijing enriched by remote participants from all around the world. However, as the delta variant of COVID was followed by omicron, and border and travel restrictions around the globe intensified rather than being removed, we made the difficult decision to move to a fully virtual ICPE for the third year in a row.

Despite this, we hope that we were able to prepare a program that is no less exciting than previous iterations of ICPE. This year, we extend a warm welcome to three excellent keynote speakers, covering a range of industrial and academic topics - Ivona Brandic (TU Vienna), John Wilkes (Google), and Longxiang Li (Inspur). Following ICPE tradition, the technical program will consist of a healthy mix of academic and industrial contributions - nine full research paper presentations, four short research paper presentations, as well as nine presentations in the industry and experience track. Additionally, the program will offer workshops, tutorials, a work-in-progress track, as well as a demo/poster track. Finally, and for the first time, we have also included a data challenge, where students and researchers were able to study a large dataset of real-life performance traces donated by MongoDB Inc.

ICPE 2022 Welcome

General Chairs

Dan Feng, Huazhong University of Science and Technology, Wuhan, China
Steffen Becker, University of Stuttgart, Germany

Program Chairs

Nikolas Herbst, University of W├╝rzburg, Germany
Philipp Leitner, Chalmers and University of Gothenburg

ICPE 2022 Conference Organization

ICPE 2022 Sponsors & Supporters

Table of contents

Keynote Talks

Data Science Driven Methods for Sustainable and Failure Tolerant Edge Systems

Authors
Ivona Brandic
DOI
10.1145/3489525.3511694
Pages
1 -- 1

Performance Optimization of HPC Applications in Large-Scale Cluster Systems

Authors
Longxiang Li
DOI
10.1145/3489525.3511696
Pages
3 -- 3

Session 1: Service and Cloud Computing

LongTale: Toward Automatic Performance Anomaly Explanation in Microservices

Authors
Richard Li
Min Du
Zheng Wang
Hyunseok Chang
Sarit Mukherjee
Eric Eide
DOI
10.1145/3489525.3511675
Pages
5 -- 16

An Empirical Study of Service Mesh Traffic Management Policies for Microservices

Authors
Mohammad Reza Saleh Sedghpour
Cristian Klein
Johan Tordsson
DOI
10.1145/3489525.3511686
Pages
17 -- 27

Best Practices for HPC Workloads on Public Cloud Platforms: A Guide for Computational Scientists to Use Public Cloud for HPC Workloads

Authors
Robert Walkup
Seetharami R. Seelam
Sophia Wen
DOI
10.1145/3489525.3511693
Pages
29 -- 35

NVMe Virtualization for Cloud Virtual Machines

Authors
Lixiang Luo
I-Hsin Chung
Seetharami Seelam
Ming-hung Chen
Yun Joon Soh
DOI
10.1145/3489525.3511688
Pages
37 -- 46

Session 2: GPUs and Heterogeneous Platforms

Exploring the Use of Novel Spatial Accelerators in Scientific Applications

Authors
Rizwan A. Ashraf
Roberto Gioiosa
DOI
10.1145/3489525.3511690
Pages
47 -- 58

Extending SYCL's Programming Paradigm with Tensor-based SIMD Abstractions

Authors
Wilson Feng
Shucai Yao
Kai Ting Wang
Md Aamir Raihan
Laichun Feng
Chunrong Xu
DOI
10.1145/3489525.3511681
Pages
59 -- 66

Oversubscribing GPU Unified Virtual Memory: Implications and Suggestions

Authors
Chuanming Shao
Jinyang Guo
Pengyu Wang
Jing Wang
Chao Li
Minyi Guo
DOI
10.1145/3489525.3511691
Pages
67 -- 75

Isolating GPU Architectural Features Using Parallelism-Aware Microbenchmarks

Authors
Rico van Stigt
Stephen Nicholas Swatman
Ana-Lucia Varbanescu
DOI
10.1145/3489525.3511673
Pages
77 -- 88

Session 3: Empirical Studies of Performance

Same, Same, but Dissimilar: Exploring Measurements for Workload Time-series Similarity

Authors
Mark Leznik
Johannes Grohmann
Nina Kliche
André Bauer
Daniel Seybold
Simon Eismann
Samuel Kounev
Jörg Domaschka
DOI
10.1145/3489525.3511699
Pages
89 -- 96

Studying the Performance Risks of Upgrading Docker Hub Images: A Case Study of WordPress

Authors
Mikael Sabuhi
Petr Musilek
Cor-Paul Bezemer
DOI
10.1145/3489525.3511683
Pages
97 -- 104

Why Is It Not Solved Yet?: Challenges for Production-Ready Autoscaling

Authors
Martin Straesser
Johannes Grohmann
Jóakim von Kistowski
Simon Eismann
André Bauer
Samuel Kounev
DOI
10.1145/3489525.3511680
Pages
105 -- 115

Evaluating the Scalability and Elasticity of Function as a Service Platform

Authors
Kim Long Ngo
Joydeep Mukherjee
Zhen Ming Jiang
Marin Litoiu
DOI
10.1145/3489525.3511682
Pages
117 -- 124

Session 4: Machine Learning and Performance

Machine Learning based Interference Modelling in Cloud-Native Applications

Authors
Alexandru Baluta
Joydeep Mukherjee
Marin Litoiu
DOI
10.1145/3489525.3511677
Pages
125 -- 132

Performance Model and Profile Guided Design of a High-Performance Session Based Recommendation Engine

Authors
Ashwin Krishnan
Manoj Nambiar
Nupur Sumeet
Sana Iqbal
DOI
10.1145/3489525.3511692
Pages
133 -- 144

The Cost of Reinforcement Learning for Game Engines: The AZ-Hive Case-study

Authors
Danilo de Goede
Duncan Kampert
Ana Lucia Varbanescu
DOI
10.1145/3489525.3511685
Pages
145 -- 152

Session 5: Hardware Performance

Alternating Blind Identification of Power Sources for Mobile SoCs

Authors
Sofiane Chetoui
Michael Chen
Abhinav Golas
Farrukh Hijaz
Adel Belouchrani
Sherief Reda
DOI
10.1145/3489525.3511676
Pages
153 -- 164

Memory Performance of AMD EPYC Rome and Intel Cascade Lake SP Server Processors

Authors
Markus Velten
Robert Schöne
Thomas Ilsche
Daniel Hackenberg
DOI
10.1145/3489525.3511689
Pages
165 -- 175

Near-Storage Processing for Solid State Drive Based Recommendation Inference with SmartSSDs®

Authors
Mohammadreza Soltaniyeh
Veronica Lagrange Moutinho Dos Reis
Matt Bryson
Xuebin Yao
Richard P. Martin
Santosh Nagarakatte
DOI
10.1145/3489525.3511672
Pages
177 -- 186

HLS_Profiler: Non-Intrusive Profiling Tool for HLS based Applications

Authors
Nupur Sumeet
Deeksha Deeksha
Manoj Nambiar
DOI
10.1145/3489525.3511684
Pages
187 -- 198

Session 6: Theory of Performance

A Mixed PS-FCFS Policy for CPU Intensive Workloads

Authors
Simonetta Balsamo
Andrea Marin
Isi Mitrani
DOI
10.1145/3489525.3511678
Pages
199 -- 210

A Stochastic Extension of Stateflow

Authors
Stefan Kaalen
Anton Hampus
Mattias Nyberg
Olle Mattsson
DOI
10.1145/3489525.3511679
Pages
211 -- 222

Sampling-based Label Propagation for Balanced Graph Partitioning

Authors
Adnan El Moussawi
Ricardo Rojas Ruiz
Nacéra Bennacer Seghouani
DOI
10.1145/3489525.3511698
Pages
223 -- 230