The Future Is in the Cards

I've wondered over the last few weeks what is in store for the humble CPU as the use of low-latency trading continues to grow. At several industry events, I've heard that everyone is trying to push as much work off the CPU as possible in order to reduce latency.
Many financial technologists are talking about adopting one of the many field-programmable gate arrays (FPGA) from a number of vendors, or graphical processing units (GPUs) from Nvidia. The biggest challenge with these alternative processors is that they use obscure and proprietary programming languages, and finding the staff with the knowledge and talent to get the most out of these platforms is hard, and they are expensive to keep.
However, there are some interesting developments on the horizon. Processor maker Intel is exploring a way to deliver additional processors on separate cards to take some of the workload off the CPU. At the same time, Microsoft is in the midst of Project Kiwi, which aims to make reconfigurable computing platforms like FPGAs more accessible to mainstream programmers.
If these vendors successfully opened up these card-based processors to a wider base of programmers, I predict their adoption would shoot through the roof. But it will be at least 12 to 18 months before we see products from the vendors.
For a long time, the release cycles of these alternative computing platforms seemed 180 degrees out of sync with the mainstream processors. When users had pushed their existing CPUs to the limit, many would turn to cards and appliances to get an added performance boost until the next generation of CPUs hit the market. With the major processors backing the new architecture, this release cycle could soon change.
Send me your thoughts on the topic at rob.daly@incisivemedia.com
Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.
To access these options, along with all other subscription benefits, please contact info@waterstechnology.com or view our subscription options here: https://subscriptions.waterstechnology.com/subscribe
You are currently unable to print this content. Please contact info@waterstechnology.com to find out more.
You are currently unable to copy this content. Please contact info@waterstechnology.com to find out more.
Copyright Infopro Digital Limited. All rights reserved.
As outlined in our terms and conditions, https://www.infopro-digital.com/terms-and-conditions/subscriptions/ (point 2.4), printing is limited to a single copy.
If you would like to purchase additional rights please email info@waterstechnology.com
Copyright Infopro Digital Limited. All rights reserved.
You may share this content using our article tools. As outlined in our terms and conditions, https://www.infopro-digital.com/terms-and-conditions/subscriptions/ (clause 2.4), an Authorised User may only make one copy of the materials for their own personal use. You must also comply with the restrictions in clause 2.5.
If you would like to purchase additional rights please email info@waterstechnology.com
More on Trading Tech
Doing a deal? Prioritize info security early
Engaging information security teams early in licensing deals can deliver better results and catch potential issues. Neglecting them can cause delays and disruption, writes Devexperts’ Heetesh Rawal in this op-ed.
Google gifts Linux, capital raised for Canton, one less CTP bid, and more
The Waters Cooler: Banks team up for open-source AI controls, S&P injects GenAI into Capital IQ, and Goldman Sachs employees get their own AI assistant in this week’s news roundup.
Waters Wavelength Ep. 323: MarketAxess’s Chowdhury and Burke (plus some Cusip updates)
This week, Riad Chowdhury, head of Asia-Pacific, and Dan Burke, global head of emerging markets at MarketAxess, join to discuss block trading in fixed income. Plus Reb discusses her recent article about Cusip and updates on the class action lawsuit moving through the courts.
As datacenter cooling issues rise, FPGAs could help
IMD Wrap: As temperatures are spiking, so too is demand for capacity related to AI applications. Max says FPGAs could help to ease the burden being forced on datacenters.
WatersTechnology latest edition
Check out our latest edition, plus more than 13 years of our best content.
Deutsche Bank casts a cautious eye towards agentic AI
“An AI worker is something that is really buildable,” says innovation and AI head
LLMs are making alternative datasets ‘fuzzy’
Waters Wrap: While large language models and generative/agentic AI offer an endless amount of opportunity, they are also exposing unforeseen risks and challenges.
Trading venues seen as easiest targets for Esma supervision
Platforms do not pose systemic risks for member states and are already subject to consistent rules.