NanoSpeed introduces Nano-Risk FPGA for CME



Author: Sanjay Shah, CTO, NanoSpeed
Date: 3 November 2014
Source: http://www.wallstreetletter.com/news/nanospeed_introduces_nano-risk_fpga_for_cme

NanoSpeed, a provider of ultra-fast FPGA solutions to the trading community, has released its latest Nano-Risk FPGA with a latency of sub 60 nanoseconds for the CME.

Sanjay Shah, CEO of NanoSpeed, said that the latency in the previous version was around 300 nanoseconds and they have reduced it to under 60 nanoseconds.

“Typically it is about 50 to 100 times faster than the banks’ software-based systems. And the reason why it is fast in the FPGA is that we are able to do a lot of these operations in parallel - all of these risk checks, there could be 30 risk checks that we would typically be doing,” he said.

“If you do them one after the other [which is typically done in software] then it can take quite a long time. But if you can do them all at the same time using parallel computational resources, then it is so much faster,” Shah told WSL.

Besides the latency, the difference between the first generation system and the new one is that it is now able to cater for the CFTC’s Concept Release on Risk Controls and System Safeguards for Automated Trading regulation and also the ESMA’s guidelines on automated trading.

“The previous version was just catering for the SEC’s 15C3-5 regulation, so now we can cater for these two other regulations as well,” said Shah.

Whereas the new version was developed for the CME, Shah said that they are having “quite a lot of customer interest in this market and are planning to roll out this product for other venues as well – for NASDAQ, Eurex, LSE, some markets in Asia and more markets in Europe”.

The latency is measured using optical measurement tools and every single message that goes through is time-stamped.

“We keep a log of that and the clients will have a full access to these trading logs. They can see the statistics – the time and the latency,” explained Shah.

Uniquely, NanoSpeed uses the aerospace development model as the basis for its design and testing.

With a latency of sub 60 nanoseconds the product is the new benchmark for fastest pre-trade risk checks.

Shah said that this latency is not possible to reduce much further: “We are getting close to the limit now.”

According to Shah, Nano-Risk is an excellent fit for proprietary trading firms, statarbs, hedge funds and SDMA brokers.

Besides the pre-trade risk checks, NanoSpeed is also looking into the microwave links between different venues to develop FPGA data compressors and fair bandwidth management.

“There are a lot of exciting products in the pipeline. We are constantly striving to get new products out to new venues” said Shah.

In terms of reliability problems with existing systems, Shah said that “they have been lucky and have not had any problems so far”.

“I think partly the reason for that is that we design things with reliability and testability in mind from the beginning. If there are any changes, our systems will automatically test themselves,” he stressed.