Arthur Petitpierre

Aperf is an open-source project created by AWS to accelerate the performance debugging process by enabling experienced users to deep dive into extensive data and less experienced users to identify issues without specialized knowledge.

Getting the best compute performance is an art, and knowing low-level details helps a lot. This talk covers everything you've wanted to know: memory topology, accessing hardware counters, hyperthreading side-effects, and how to properly run performance tests and optimize your latency for production.

In the early days of Amazon EC2, processors were simple single-socket, single-die designs. Today's processors have evolved into complex architectures featuring multiple dies, multiple sockets, and intricate memory hierarchies—developments that profoundly shape how we architect applications and tune their performance.

Many developers are working at a higher and higher level, which is fantastic from a productivity standpoint. However, when it comes to performance, it is often necessary to be closer to the metal to understand the dynamics of applications, study their behavior, and understand and fix their bottlenecks.

The availability of processors with and without hyperthreading makes performance evaluation a tricky game. In this code talk, study a web application and evaluate its performance in various scenarios, and discover how to optimize throughput and latency along the way.

When asked to compare performance at the infrastructure level, many people are running benchmarks. Most of the time, benchmarks aren’t representative of their applications’ behaviors.

Using processors instead of GPUs can be very cost-effective for some ML workloads. In this chalk talk, learn how AWS Graviton3 can accelerate your inference workloads even more.

Many developers are working at a higher and higher level, which is fantastic from a productivity standpoint. However, when it comes to performance, it is often necessary to be closer to the metal to understand the dynamics of applications, study their behavior, and understand and fix their bottlenecks.

This chalk talk details best practices and lessons learned about coding for multiple architectures (x86 and AWS Graviton) from the last ten years in the high performance computing (HPC) community. Amazon EC2 has seen a significant rise in the adoption of Arm-based AWS Graviton instances.

AWS Graviton processors feature key capabilities that enable you to run cloud-native applications securely and at scale. AWS Graviton3 processors feature always-on memory encryption, dedicated caches for every vCPU, and support for pointer authentication.

AWS Graviton processors are designed to deliver the best price performance for a wide variety of workloads in Amazon EC2. In this workshop, walk through moving a workload to AWS Graviton–based instances, including containerized applications.

AWS Graviton processors are custom designed to enable the best price performance for a wide variety of workloads on AWS including application servers, microservices, open source–based databases and in-memory caches, gaming, high-performance computing, and machine learning. Tens of thousands of organizations are running production workloads on AWS Graviton.

AWS Graviton2 processors enable the best price performance for a wide variety of workloads in Amazon EC2. In this workshop, walk through moving a workload to AWS Graviton2-based instances, including containerized applications.

AWS Graviton2-based Amazon EC2 instances enable great price-performance for a wide variety of workloads in Amazon EC2. In this chalk talk, discover stories from customers who have successfully adopted Graviton2-based instances, including details of workloads, technologies, price-performance benefits, effort, and learnings.

AWS Graviton2 processors are custom built by AWS to enable improved price-performance for workloads in Amazon EC2. Instances powered by Graviton2 processors can provide up to 40 percent better price-performance over comparable x86-based instances for a wide variety of workloads.

In this chalk talk, learn how easy it can be to build and run various open-source applications on Amazon EC2 A1 Arm-based instances while lowering your costs. We discuss the software building blocks and toolchain support, as well as walk through example web, containerized application, and reference architectures that you can build on the Arm platform.

Amazon EC2 A1 instances are the first EC2 instances powered by Arm-based AWS Graviton processors. They deliver significant cost savings for scale-out and Arm-based applications, such as web servers and containerized microservices.