AWS re:Invent

EC2 F1 Sessions:

Accelerating Development Using Custom Hardware Accelerations with Amazon EC2 F1 Instances

Date: Monday, November 26 | Tuesday, November 27
Time: 2:30 PM - 3:30PM | 10:45 AM - 11:45 AM
Presenter: Kristopher King, Gadi Hutt, Plout Galatsopoulos
Amazon EC2 F1 instances with field programmable gate arrays (FPGAs), combined with optimized cloud-based FPGA development tools, provides researchers, application developers, and startups with a well-tested, standardized, and accessible platform for custom hardware-accelerated computing. In this session we will dive deep on how to optimize acceleration development on AWS, and we will hear from our guest speaker on how they're enabling new silicon engineering capabilities using EC2 F1 instances.

Accelerate Your C/C++ Applications with Amazon EC2 F1 Instances

Date: Wednesday, November 28
Time: 2:30 PM - 4:45PM
Presenter: Kristopher King
Amazon EC2 F1 OpenCL development workflow helps software developers with little to no FPGA experience to supercharge their applications with Amazon EC2 F1. Join us for an overview and demonstration of accelerating your C/C++ applications in the cloud using OpenCL with Amazon EC2 F1 instances. In this workshop, we walk through the development flow for creating a custom hardware acceleration for a software algorithm. Attendees get hands-on and creative by optimizing an algorithm for maximum acceleration on Amazon EC2 F1 instances. All attendees must bring their own laptop (Windows, macOS, and Linux all supported). Tablets are not appropriate. We also recommend having the current version of Chrome or Firefox installed.

Integrate the AWS Cloud with Responsive Xilinx Machine Learning at the Edge

Date: Wednesday, November 28
Time: 6:15 PM - 8:30 PM
Presenter: Wesley Skeffington, Graham Schelle, Richard Elberger
In this workshop, learn how you can integrate Xilinx FPGA SoC based machine learning with massive scale AWS IoT Analytics.  Based on an oil & gas product scenario, you will learn how to combine AWS Cloud services with AWS Greengrass on  Zynq Ultrascale+  and Amazon FreeRTOS on Xilinx Zynq-7000.  After this workshop, you will have a concrete understanding of Machine Learning applicability at the edge and its relationship with the AWS Cloud and Xilinx edge platforms.

Booth Demos:

Booth Information

Xilinx Booth #1606
Venetian Hotel
Las Vegas, NV

Monday, November 26
4:00 PM - 7:00 PM

Tuesday, November 27
11:00 AM - 6:00 PM

Wednesday, November 28
11:00 AM - 6:00 PM

Thursday, November 29
11:00 AM - 4:00 PM

Demo Title Description
a:FreeRTOS on Xilinx MicroZed SoC Platform
  • Fully qualified AWS FreeRTOS running on the Avnet MicroZed platform built on the Xilinx Zynq 7000 FPGA SoC.
  • The platform demonstration shows an industrial intelligent I/O module which combines the AWS IoT software stack running on the ARM A9 processing system working in collaboration with the FPGA fabric. 
  • In the demonstration the platform not only send raw sensor data to the AWS IoT Analytic service but also uses the AWS device shadow to synchronize the state of the device to trigger a remote alarm but to also provide a mechanism for a system update.
AWS Greengrass on Xilinx Zync Ultrascale+, Ultra96, and ZCU104 SoC Platforms
  • The powerful AWS Greengrass SDK will be shown operating on two new Xilinx Zynq Ultrascale+ platforms in the Avnet Ultra96 and the Xilinx ZCU104 platforms. 
  • The ZCU104 will be showing a SageMaker and Greengrass deployed edge machine learning inference solution that implements person detection for securing a remote industrial machine response running in parallel to an advanced high-speed motor control application in less than 8W; while the Ultra96 will be showing an edge Lambda function ingesting data from the FPGA, filtering the data, and sending an event based data stream to AWS IoT Core.
Highest Quality Real-Time VP9 Encoding
  • Highest quality 1080p60 real-time VP9 encoder
  • Integrated Adaptive Bit Rate (ABR) capability for live streaming applications
  • Easy to integrate using FFmpeg
  • Adaptable due to inherent programmable nature of FPGA’s for future proofing
Apps & Libraries for F1