NovaSparks, a market data processing company-based on field programmable gate arrays (FPGAs), released audited performance benchmarks that verify the vendor’s ability to process market data at deterministic speeds of 1.4 microseconds.
The results set a new record for deterministic market data processing, said the company in yesterday’s release.
The 1.4 microseconds mean latency rate is maintained when the market data is accelerated 20 times the actual market speed, demonstrating the deterministic nature of the NovaSparks Market Data Matrix, the company said. The test conditions equate to a millisecond peak rate of 18.9 million messages per second.
In mathematics and physics, a deterministic system is one in which no randomness is involved in the development of future states of the system. A deterministic model will always produce the same output from a given starting condition or initial state.
“Deterministic processing of market data at ultra-low latency rates is a breakthrough for an industry that is constantly re-assessing their ability to trade across all market conditions, “ stated Michal Sanak, CIO at RSJ, a proprietary trading firm that is listed as the largest NYSE Liffe trader.
The accelerated data was replayed from NASDAQ TotalView ITCH 4.1 feed during the open and close periods of the day, periods that are considered the most challenging time of day to maintain a full depth of book at deterministic speed.
“The tests rigorously measure ‘wire-to-memory’ latency, leaving nowhere to hide,” stated Peter Lankford, director of the Securities Technology Analysis Center(STAC), an independent body that conducted the tests with a detailed latency benchmark, called STAC-M, which is available to the public. The benchmark includes all the tasks to receive, build and maintain a full depth of book starting at the time the first byte of the packet is received by the system. The specification provides a reproducible framework for measuring the performance of traditional software-based solutions as well as pure FPGA solutions like the NovaSparks Market Data Matrix.
“As an industry, we need to cut back on the amount of duplicative testing that takes place and start to share more best practices,” according to Olivier Baetz, COO of NovaSparks North America in the firm’s release. “We believe that full performance transparency will help trading firms determine the business case for deterministic market data processing,” stated Baetz.
NovaSparks offers the Market Data Matrix for 27 feeds. The feeds, covering major equities and futures exchanges, exhibit similar deterministic microsecond speeds. The Market Data Matrix feed handlers include a distribution layer in the FPGA architecture that allows direct connection of up to 21 trading servers without any loss of performance.
Ivy is Editor-at-Large for Advanced Trading and Wall Street & Technology. Ivy is responsible for writing in-depth feature articles, daily blogs and news articles with a focus on automated trading in the capital markets. As an industry expert, Ivy has reported on a myriad ... View Full Bio