By Steve Ranger, 7 April 2006 12:30
NEWS
Companies will have to rewrite some of their legacy code if they want to take full advantage of the move to multi-core computing.
Software code designed to run on single-core systems will have to be modified to make the most of future multi-core systems, according to consultancy Accenture, which is developing tools to make this process easier.
Martin Illsley, director of research at Accenture Technology Labs, told silicon.com: "You can't run your old code and expect it to run faster. There's a lot of very old code out there so there will be quite a disruption to legacy code to take advantage of multi-core architecture."
Mike Redding, director of development at the labs, added: "What we are working on is a diagnostic to say [whether] there is a concern for this piece of code and how important is it. We are researching how to identify the code, to put a price on the cost-benefit of switching."
Redding said that as their legacy code ages, companies will have to chose between the cost of more hardware and the cost of modifying the code.
"We are trying to identify which code has the business case," he said.
But he added: "It's not a looming crisis like Y2K - it's not going to be a Y2K bonanza. There's going to be good steady work."
Comments
There are 2 comments. Join the discussion
1. Kunel Patel
Legacy and Bespoke applications do not necessarily need to be re-written for dual core systems. The need and requirement for this will be driven by the business and its functional requirements.
Thus, a focus needs to be placed on performance and the value chain around the application. If the end client has an application which would have limited increase in performance from a dual-proc system than there is little point in investing in a re-write or upgrade. Candidates which come to mind would be front end applications which merely serve as a gateway to back-end solutions.
Seems a lower level priority in enterprise, and I beleive a larger focus needs to be placed on emerging and developing applications.....
2. Simon
"You can't run your old code and expect it to run faster"
As they say around here, "no s**t Sherlock !" About as hard to work out as "Ice cubes don't store well in the oven".
Perhaps if it's legacy code, it doesn't actually NEED the extra performance, or perhaps doesn't justify the cost of a re-write.
But then, what do you expect from a salesman "I have a wonderful tool here, but you actually don't need it at all" ?