One of the biggest challenges you might face if you want to start working with Alveo either on premises or in the cloud is to identify the one which is best suited to your application.
With several cards to chose from it can be a daunting task, but it does not need to be. There are several factors we can use to help us identify the most appropriate Alveo card, these selection criteria include application space, logic resources / performance, memory requirements, interfacing capabilities and of course thermal load.
Alveo cards can be grouped as to provide support for one of four distinct application spaces
Video Streaming –Cards in this range focuses on, low and ultra low video processing supporting standards such as H264, H265 and Adaptive Bit Rate Scaling. Cards in this range are also able to perform faster than real time processing when transcoding videos, along with supporting supporting live events and cloud gaming.
Data Centre Infrastructure – This range provides network acceleration and computational storage enabling developers to be accelerate data centre infrastructure. This enables the integration of IP Security, and technologies such as Open vSwitch which enables network switching for virtualised environments.
Compute – Compute cards provide developers with a range of resources which can be used for the deployment of compute intensive applications. Examples of this include Database and Data Analytics, Machine Learning and of course data security and compression.
Financial Technology – Cards in this group are designed to enable low latency electronic trading, along with data and risk analytics.
Of course within a application space we need to determine how best to select a specific card. In this case we can leverage parameters such as logic resources and performance, high performance memories and interfacing solution. Another key selection criteria is the development methodology, as while most Alveo cards can be developed using Vitir or Vivado depending upon your entry point. Some cards such as the those in the Financial technology application space are programmed using Vivado only.
When it comes to choosing between Video Processing Cards the developer is able to be guided by the performance requirement. If ultra-low latency is required or support for the next generation AV1 compression standard (in addition to H.264 and H.265), the Alveo MA35D is selected. This card is unique in that it contains a custom ASIC and not programmable logic hence it is programmed using the Advanced Media Acceleration SDK. If a low latency approach is required and support for H.264 and H.265 codecs necessary, the Alveo U30 is a good choice which can leverage the Video SDK for development.
The largest possible selection range is within the compute application space. Within this space we can use the size of the module (single or double) as a good initial guide. Once we have decided on the size of the card are can then select depending upon if Cache Coherent Interface for Accelerators (CCIX) is needed. If CCIX is required this drives to a U50 depending upon the slot size. If the application demands the highest bandwidth memory access a device which provides HBM is required again this identifies the U50. If neither CCIX or HBM is required we can select the Alveo compute accelerator card based purely of the resources required for the application and future growth.
Selecting the network acceleration card depends upon the line rate required for the network inface if speed above 25Gb Ethernet the Alveo U45N is required. If line rates are below 25Gb the selection criteria becomes again one of resources and supported capabilities.
The final application space of Fintech requires the developer to select between functionality of low latency network acceleration or if adaptable and programmable resources are also required.
What ever card we select in our next blog we are going to get started, looking at how to create designs using Vivado on the Alveo U55C.
Workshops and Webinars If you enjoyed the blog why not take a look at the free webinars, workshops and training courses we have created over the years. Highlights include
Professional PYNQ Learn how to use PYNQ in your developments
Introduction to Vivado learn how to use AMD Vivado
Ultra96, MiniZed & ZU1 three day course looking at HW, SW and Petalinux
Arty Z7-20 Class looking at HW, SW and Petalinux
Mastering MicroBlaze learn how to create MicroBlaze solutions
HLS Hero Workshop learn how to create High Level Synthesis based solutions
Perfecting Petalinux learn how to create and work with petalinux OS
Embedded System Book
Do you want to know more about designing embedded systems from scratch? Check out our book on creating embedded systems. This book will walk you through all the stages of requirements, architecture, component selection, schematics, layout, and FPGA / software design. We designed and manufactured the board at the heart of the book! The schematics and layout are available in Altium here Learn more about the board (see previous blogs on Bring up, DDR validation, USB, Sensors) and view the schematics here.
Order here Sponsored by AMD
Comments