Supercomputing 2022, the leading technical program in the high-performance computing community, is right around the corner. This year, Achronix is excited to be presenting and exhibiting our computing technology throughout the conference.
We are the industry’s only independent supplier of high-end FPGAs and embedded FPGA (eFPGA) IP solutions, and we have some important technology advancements to share with the HPC community. Read on to get a sneak peek of some of the exciting topics we’ll be presenting at this year’s conference and five reasons why you should join us:
1: It’s the Best HPC Show Globally
Started in 1988, the Supercomputing Conference is recognized as the leading forum where the world’s foremost high-performance computing (HPC) experts come together to share ideas and inventions that could pave the path for a better future.
HPC is, by definition, all about high performance, and Achronix FPGAs and eFPGA IP are designed for high performance. Between our 2D NoC, MLPs, GDDR memory, and more, our solutions result in faster computation and reduced time to market.
2: Beer & Popcorn Happy Hour with Special Presentations
From 3:30 pm – 5:00 pm CST on Tuesday, November 15th, we’ll be hosting a beer and popcorn happy hour in Booth #4123. During the happy hour, we’ll be giving the following short technical presentations (plus hosting an exciting raffle with giveaways you won't want to miss):
- Use of High-Performance FPGA Chiplets in Heterogeneous Systems: As chiplets become more prominent in the field of high-performance computing, what role do FPGAs play? In this talk, we’ll discuss the many use cases and benefits of FPGA chiplets in heterogeneous systems, and why this technology could pave the path for the future of HPC.
- Unlocking the Full Potential of Conversational AI with FPGAs: An FPGA can be a very attractive platform for many machine learning (ML) inference applications. In this presentation, using the example of automatic speech recognition (ASR), we will explore how FPGAs can be used to accelerate the conversational AI and natural language processing applications. We’ll review the key components in the Achronix FPGA architecture, such as the 2D network on chip (NoC), high-speed external GDDR6 memory interfaces, and optimized machine learning processors (MLPs), and how our symmetrical architecture can make scaling solutions in the FPGA transparent and rapid to deploy. Using standard benchmarks, we demonstrate an ASR appliance that can reduce costs by as much as 90% when compared with alternative approaches.
- Accelerating SmartNIC Performance with Achronix Network Infrastructure Code (ANIC): SmartNICs now dominate the high-performance Ethernet market, enabling customers to offload and accelerate network functionality and free up key CPU cycles for computation. With an ANIC, customers can utilize a pre-built, selectable SmartNIC pipeline on an Achronix VectorPath PCIe card at 100GbE line rates and accelerate network functions on a high performance, fully programmable Speedster7t FPGA. We will review the key elements of the ANIC architecture, how customers can choose which components are necessary for their needs, and how their network application can be added into the SmartNIC pipeline for accelerated performance.
3. Exhibitor Forum Presentation
At the Exhibitor Forum (located D171) on Thursday, November 17th at 4:30 pm CST, Robert Ellison, North America Product Lead at Achronix, will give a special presentation, entitled “Integration of High-Performance FPGA Chiplets in Heterogeneous Systems”.
There is a growing need to closely couple heterogeneous computing elements within a high-performance computing platform in order to combat the increasing development costs of monolithic solutions. For the world of FPGAs, this has meant the rise of eFPGAs which allow for tighter coupling of an FPGA fabric with a CPU core.
However, modern advances and standardization in die-to-die interfaces have ushered in a new era where chiplets can enable even tighter coupling than ever before. FPGA chiplets are designed to be mixed and matched with other chiplets performing diverse functions to enable higher performance, lower development costs, and more scalable FPGA solutions than ever before.
This talk will examine the UCIe open-standard, FPGA architecture as a standardized interface, and the benefits of closely coupling FPGA accelerators with heterogeneous computing elements within a high-performance computing platform.
4: High-Performance Product Demonstrations
At our booth (#4123) we’ll have demos of our FPGA and high-performance products. Specifically, we’ll be demonstrating automatic speech recognition and fully programmable FPGA SmartNICs on our platforms and showing how our FPGA solutions can serve as an ideal hardware accelerator for even the most complex computing tasks.
5: It’s All Free
If all of that wasn’t persuasive enough, it’s also all free! Anyone can request a free exhibitor pass for SC22 here.
Come find us at Supercomputing! We look forward to meeting you in Dallas!