RiTC 3 : Runtimes in the Cloud 2020
HPCA 2020
About the Tutorial
Modern data-centers must support sprawling, interconnected software systems subject to constant revision. The complexity of software continuous evolution, deployment and interactions makes the exact runtime conditions of software difficult to predict and as consequence, difficult to optimize. Cloud computing exacerbates the complexity of this problem by expanding the diversity of codes potentially involved in these situations.
In light of this growing software complexity, maximizing system performance requires runtime mechanisms to continuously measure workload behavior and adapt to dynamic workload conditions. In this tutorial, speakers from industry and academia will examine how they combine hardware and runtime systems such as Java, .Net and other emerging language runtime to meet performance needs today, while highlighting where today’s systems fall short.
Tutorial Format
The full day tutorial will be composed of lectures, a hands on lab and a panel.
- Lectures from companies from the industry and academia on the following topics:
- Optimizing Runtime Workloads on Traditional and New Architectures
- Hardware Assisted Runtime Systems
- Runtime Optimization at Scale
- Code generation optimization for modern hardware
- Lectures on Research topics in Cloud Runtimes from academia
- Hands on Lab on the Cloud
- Panel Discussion (speakers from the industry and academia)
Target Audience
Researchers that are advocating new ways of solving runtime problems in hardware and software, particularly as it relates to the cloud.
Practitioners that need to solve runtime performance problems in the cloud.
Schedule
The tutorial will be held in the Marseilles Room San Diego Mission Bay Resort on Sunday February 23rd 2020
9:00a – 9:15a
Opening Remarks
9:15a – 10:00a
Bridging the Gap Between Deep Learning Models and “the Metal” with Apache TVM & VTA [slides]
Luis Ceze (U. of Washington)
10:00a – 10:30a
Break
10:30a – 11:15a
HHVM Jumpstart: towards instant JIT warmup and agile cluster operations [slides]
Bin Liu (Facebook)
11:15a – 12:00p
Hardening your Runtime [slides]
Rodolph Perfetta (Arm)
12:00p – 1:00p
Lunch
1:00p – 1:45p
The Performance Journey of Serverless Azure Functions [slides]
Mei-chin Tsai and Hamid Safi (Microsoft)
1:45p – 2:30p
Migrating AWS Lambda’s front end from Java 8 to Java 11
Yishai Galatzer
2:30p – 3:00p
Break
3:00p – 4:00p
Tutorial – FaaS and CPU Architecture [slides]
Suresh Srinivas (Intel)
Andrea Pellegrini (Arm)
Prerequisite knowledge
College level knowledge of computer architecture, system performance, and runtime systems (e.g. Java)
Basic familiarity with Cloud and Performance Tools (e.g. Linux perf)
Tutorial Begins in
Day(s)
:
Hour(s)
:
Minute(s)
:
Second(s)