AMD's new FPGA accelerator card aims to offer stock exchanges and professional traders, both in the traditional and crypto sectors, a much more powerful, faster and more reliable tool for their trading operations.
Renowned semiconductor company AMD recently introduced its AMD Alveo™ UL3524 accelerator card. This is a new fintech accelerator designed for ultra-low latency trading applications and powered by artificial intelligence. The Alveo UL3524 is powered by a 16nm Virtex™ UltraScale+ FPGA chip and features a novel architecture with optimized and reinforced network connectivity cores for breakthrough performance. Thanks to these advancements, the AMD Alveo™ UL3524 offers a 7x improvement in latency over previous generation FPGA technology.
In short, this new AMD Alveo accelerator card provides commercial companies and professional traders with nanosecond trading execution performance, all supported by a system of trading strategies supported and optimized by artificial intelligence (AI).
Hamid Salehi, AMD's director of product marketing, commented following the launch:
In ultra-low latency trading, a nanosecond can determine the difference between a profitable or losing trade. The Alveo UL3524 accelerator card is powered by AMD’s lowest latency FPGA transceiver, specifically designed to give our fintech customers an unprecedented competitive advantage in the financial markets.
High-powered hardware supported by AI
Featuring 64 ultra-low latency transceivers, 780.000 LUTs (Look Up Tables, the basic logical unit of an FPGA) and 1.680 DSP compute slices, the Alveo UL3254 is designed to accelerate custom trading algorithms in hardware, where traders can adapt their design to evolving market conditions and strategies.
Supported by traditional FPGA flows using Vivado™ Design Suite, the product comes with a set of reference designs and performance benchmarks that enable FPGA designers to quickly explore key metrics and develop customized business strategies based on specifications, backed by global support from AMD domain experts. All of this development has been made possible by AMD’s recent acquisition of Xilinx, making it the undisputed leader in the FPGA industry.
Best of all, this accelerator is supported by open source AI technology. To do this, AMD offers developers the FINN framework. Thanks to this, you can use PyTorch and neural network quantization techniques to create fast, high-performance AI models, compile them, integrate the network model into the algorithm's data path for low-latency performance. This way, you can create trading strategies that you can implement and accelerate the operational decision set, using this accelerator and its AI capabilities.
A major evolution by AMD
The arrival of the Alveo UL3524 and purpose-built FPGA technology is enabling AMD’s strategic partners to create custom infrastructure and solutions for the fintech market. Partner solutions currently available include offerings from Alpha Data and Hypertec.
In the case of Alpha Data, its new product, the ADA-R9100 has been one of the first to use AMD's new FPGA. David Miller, CEO of Alpha Data, had this to say about the adoption of this new FPGA.
AMD’s new Virtex UltraScale+ FPGA is a game-changer for ultra-low latency networking and commerce. We have developed the ADA-R9100 rack-mount device that enables customers to easily access the full potential of the new AMD FPGA device.
For its part, Hypertec Hypertec, a well-known company that develops hardware and software solutions for high-frequency trading and specialized blockchain infrastructure, has designed a new product under the name of ORION HF X410R-G6. David Lim, Product Marketing Manager at Hypertec, said:
Hypertec engineers specifically designed the HF X410R-G6 to take full advantage of the capabilities and speed of the Alveo UL3524 platform, tailoring our solution to the most demanding low-latency tasks.
Continue reading: Polygon Network and De.Fi join forces to strengthen security on Web3
