11:45 AM
Low-Latency Technology Outpacing Programmers' Capabilities
Though Gordon Moore's 1965 prediction that the density of computer components would double every two years has been accurate, and computer processing speed has followed basically the same upward trajectory, the technologies' evolution is not being met with similar advances in computer programming expertise. While Intel and AMD were happy to roll out dual-core processors (followed now by quad-core and even dual quad-core processors), the programming expertise needed to truly take advantage of parallel processors is extremely hard to find in the marketplace, according to many of the speakers and attendees at WS&T's Accelerating Wall Street conference in New York in early May.
Again and again, executives said that finding enough programmers who are able to write "parallel" code -- programs that efficiently divide workloads across distributed processors -- is almost impossible. As Wall Street firms rely on multicore processing and even distributed computing to handle the ever-growing number of trade-related messages that are sensitive to any increase in data latency, the divergence between the capabilities of the technology and the capabilities of the programmers is becoming painfully evident.
No matter how many messages a powerful multicore processor can digest in a test lab, the experts said, if the software handling the message flow is written for a single-core environment, much of the value of the multicore processor is lost. That's a pretty scary statement, considering that financial firms will spend approximately $300 million just on technology to lower latency this year, noted Bob Iati, research director at TABB Group, in his presentation at the conference. A good portion of that dollar figure will be spent on multicore processors, as firms replace single-core chips. By 2010, Iati said, 49 percent of chipsets will be dual-core and 48 percent will be quad-core.
That's not to say trading desks aren't seeing vast improvements in data latency as firms continually upgrade processing capability. But if software was written to take full advantage of multicore processing capabilities, a firm's advanced trading strategy could jump ahead of competitors in the low-latency arms race.
Although I don't know where the talent will come from, history has shown that Wall Street will find the needed expertise somewhere. After all, as algorithmic trading took off, the Street managed to attract (and continues to woo) the best and the brightest quantitative analysts from the world's top universities.
Greg MacSweeney is editorial director of InformationWeek Financial Services, whose brands include Wall Street & Technology, Bank Systems & Technology, Advanced Trading, and Insurance & Technology. View Full Bio