profile

Marcos Ortiz

AWS Graviton Weekly # 40

Published over 2 years ago • 6 min read

Issue # 40: June 2nd, 2023 to June 9th, 2023

​[Read the browser version right here]​

​

Hey Reader.

Welcome to Issue # 40 of AWS Graviton Weekly, which will be focused on sharing everything that happened in the past week related to AWS Silicon: from June 2nd, 2023 to June 9th, 2023.

Before start sharing all the good stuff about Graviton, I wanted to thank Vytautas Klova and the whole CAST.AI team for the incredible conversation we had this week, which is a very special week if you are a Kubernetes user: the first commit of the project was on June 6th, 2014.

​Happy Birthday, Kubernetes.

The team from CAST.AI has built the leading all-in-one platform for Kubernetes automation, optimization, security, and cost management.

So, if you are using Kubernetes in any major cloud provider; I strongly recommend chatting with this team. You could save a ton of your cloud bills from your Kubernetes setup.

Now, back to business with Graviton stuff.

In this issue you will find:

  • Amazon announces AWS Snowblade for JWCC customers, the first device to meet U.S. Military Ruggedization Standards (MIL-STD-810H)
  • How to build a high-performance model with PyTorch 2.0 and AWS Graviton
  • The AWS Made Easy Livestream - Episode 52 - Graviton Part 2 with Arthur Petitepierre​
  • Amazon EC2 High Memory instances are now available in the US West (Oregon) and Asia Pacific (Hyderabad) regions
  • Mobiuspace case study
  • Read how Adevinta provided ARM nodes to 4k engineers
  • The keynote from Kevin Miller (VP of Amazon S3 at AWS) at the AWS Summit Berlin
  • "Freedom to Innovate: The ARM Neoverse Advantage": the keynote at the COMPUTEX 2023 event by Mohamed Awad
  • Learn how your startup could Win $150k of Arm Technology Credit​
    ​

NEWS

​Announcing AWS Snowblade for U.S. Department of Defense JWCC customers​

AWS announces the availability of AWS Snowblade for the U.S. Department of Defense (DoD) Joint Warfighting Cloud Capability (JWCC) contract customers.
AWS Snowblade is designed to provide AWS compute, storage, and other hybrid services in remote locations, including Denied, Disrupted, Intermittent, and Limited (DDIL) environments for the DoD. AWS Snowblade is the first AWS Snow Family device designed to meet U.S. Military Ruggedization Standards (MIL-STD-810H), enabling JWCC defense customers to run their operations in edge locations that can be subject to extreme temperatures, vibrations, and shocks. With support for 208 vCPU in a portable, compact 5U, half-rack width form-factor, AWS Snowblade is the densest compute device of the AWS Snow Family allowing JWCC customers to run demanding workloads in space, weight, and power (SWaP) constrained edge locations.

Learn more:

​https://aws.amazon.com/about-aws/whats-new/2023/06/aws-snowblade-us-defense-jwcc-customers/​

​Amazon EC2 High Memory instances now available in new regions​

Starting today, Amazon EC2 High Memory instances​
with 24TiB of memory (u-24tb1.112xlarge) are now available in US West (Oregon) Region and instances with 9TiB of memory (u-9tb1.112xlarge) are now available in the Asia Pacific (Hyderabad) Region. Customers can start using these new High Memory instances with On Demand and Savings Plan purchase options.

Learn more:

​https://aws.amazon.com/ec2/instance-types/high-memory/​

​https://www.sap.com/dmc/exp/2014-09-02-hana-hardware/enEN/#/solutions?filters=v:deCertified;ve:23​

​https://docs.aws.amazon.com/sap/latest/sap-hana/migrating-hana-to-hm.html​

​

Other news

​

ARTICLES AND TUTORIALS

​Build high-performance ML models using PyTorch 2.0 on AWS – Part 1, by Kanwaljit Khurmi (Principal Solutions Architect at AWS), Mike Schneider (Systems Developer at AWS), and Lai Wei (Senior Software Engineer at AWS)

This post demonstrates the performance and ease of running large-scale, high-performance distributed ML model training and deployment using PyTorch 2.0 on AWS. This post further walks through a step-by-step implementation of fine-tuning a RoBERTa (Robustly Optimized BERT Pretraining Approach) model for sentiment analysis using AWS Deep Learning AMIs
(AWS DLAMI) and AWS Deep Learning Containers (DLCs) on Amazon Elastic Compute Cloud (Amazon EC2 p4d.24xlarge) with an observed 42% speedup when used with PyTorch 2.0 torch.compile + bf16 + fused AdamW. The fine-tuned model is then deployed on AWS Graviton-based C7g EC2 instance on Amazon SageMaker with an observed 10% speedup compared to PyTorch 1.13.

Learn more:

​https://pytorch.org/blog/pytorch-2.0-release/​

​https://aws.amazon.com/machine-learning/amis/​

​https://aws.amazon.com/blogs/machine-learning/optimized-pytorch-2-0-inference-with-aws-graviton-processors/​

​

​Mobiuspace Delivers up to 40% Improved Price-Performance Using Amazon EMR on EKS​

With Amazon EMR on EKS and the ARM-based AWS Graviton 2 instances, we improved the overall performance of our big data operations by 30% and reduced costs by 20%.

​Li Rui (Vice President of Technology at Mobiuspace)

​

​Transparently providing ARM nodes to 4,000 engineers, by Thibault Jamet (Technical Product Owner at Adevinta) and Miguel Bernabeu (Site Reliability Engineer at Adevinta)

One way we reduce computation costs in the current AWS landscape is to use Graviton instances using the ARM CPU architecture. In this article, we explain how, in SCHIP, we provide a best-in-class experience to run ARM workloads alongside the x86 architecture to more than 4,000 developers. They can then reduce their infrastructure costs as transparently and gradually as possible.

Learn more:

​https://github.com/adevinta/noe​

​

​Cost-effective bulk processing with Amazon DynamoDB, by Jason Hunter (Principal Solutions Architect at AWS specializing in DynamoDB)

Your Amazon DynamoDB table might store millions, billions, or even trillions of items. If you ever need to perform a bulk update action against items in a large table, it’s important to consider the cost. In this post, I show you three techniques for cost-effective in-place bulk processing with DynamoDB.

Learn more:

​https://aws.amazon.com/blogs/database/cost-effective-bulk-processing-with-amazon-dynamodb/​

​

SLIDES, VIDEOS, and AUDIO

​[Livestream] AWS Made Easy Livestream -Ep 52 - Graviton Part 2 with Arthur Petitepierre (AWS), Rahul Subramaniam, and Stephen J. Barr from CloudFix​

​

​[VIDEO] AWS Summit Berlin 2023 Keynote​

​Kevin Miller (Vice President for Amazon S3) shared a lot of great stuff related to EC2, Storage, and more; including of course AWS Graviton

​[VIDEO] Enabling Scalable AUTOSAR Development & Testing In The Cloud - AWS Auto and Manufacturing Meetup, with Wolfgang Thieme (Head of Product Management System and Cloud Solutions at Elektrobit)

Ever-increasing software projects, complex architectures, various contributors, and increasing expectations on time to market are a challenge in the traditionally very embedded focused development workflows of the automotive industry. Having everything a automotive developer needs available right away, in a collaborative and scalable fashion everywhere you go is a game changer. Elektrobit & AWS teamed up to bring Elektrobits class-leading automotive software components together with AWS cloud services to build a new generation of tooling. Using AWS Graviton ARM-based instances to virtualize the ECU (electronic control units) in the cloud massively improves the development speed.

​

JOBS

​

EVENTS

​Graviton Essentials - Virtual Developer Day (Wednesday, July 12 2023 | 9:00 AM - 5:00 PM PDT) Live Virtual & Interactive

​

​[AWS Summit Toronto] What's new in Amazon EC2, with Pierre Tessier (Honeycomb) and Martin Yip (AWS) (June 14 from 4:00 p.m.—5:00 p.m.)

Amazon EC2 provides secure, resizable compute capacity in the cloud and makes web-scale computing easier. It is a foundational service for AWS and offers a wide variety of compute instances that are well suited for virtually every use case, from static websites to on-demand supercomputing, and it’s available with flexible pricing options. This session provides an overview of what’s new in the Amazon EC2 portfolio, including updates on capabilities, instance families, storage and networking functionality, and edge and hybrid offerings. Come learn how Honeycomb.io is leveraging AWS Graviton–based instances to reduce latency at lower costs and with fewer instances

From the ARM Ecosystem

​

​

​

Marcos Ortiz

I'm a Data Engineer by day at Riot Games (via X-Team ) and by night, I curate the last news/product announcements/resources about AWS Silicon (AWS Graviton, Nitro, Inferentia, and Trainium).

Read more from Marcos Ortiz
AWS Graviton Weekly # 139

Issue # 139: September 26, 2025, to October 10, 2025 Brought to you by CAST AI Hey Reader. Welcome to Issue # 139 of AWS Graviton Weekly, which will be focused on sharing everything that happened in the past week related to AWS Silicon: from September 26, 2025, to October 10, 2025. Enjoy Recommendation of the week: CAST AI: Move Any Workloads Without Disruption Migrate live Kubernetes containers — including those running stateful workloads — with zero downtime. Cast AI’s automation ensures...

7 days ago • 1 min read
AWS Graviton Weekly # 138

Issue # 138: September 19, 2025, to September 26, 2025 Brought to you by CAST AI Hey Reader. Welcome to Issue # 138 of AWS Graviton Weekly, which will be focused on sharing everything that happened in the past week related to AWS Silicon: from September 19, 2025 to September 26, 2025. Enjoy Recommendation of the week: CAST AI: Move Any Workloads Without Disruption Migrate live Kubernetes containers — including those running stateful workloads — with zero downtime. Cast AI’s automation ensures...

21 days ago • 1 min read
AWS Graviton Weekly 137

Issue # 137: September 12, 2025 to September 19, 2025 Brought to you by CAST AI Hey Reader. Welcome to Issue # 137 of AWS Graviton Weekly, which will be focused on sharing everything that happened in the past week related to AWS Silicon: from September 12, 2025, to September 19, 2025. Enjoy Recommendation of the week: CAST AI: Move Any Workloads Without Disruption Migrate live Kubernetes containers — including those running stateful workloads — with zero downtime. Cast AI’s automation ensures...

28 days ago • 1 min read