In a surprise move, Bluespec, developer of the only ESL synthesis toolset for control logic and complex datapaths in chip design, announced today that it has abandoned high-level software languages for its internal development of tools and has switched, instead, to assembly language.
Apparently, Bluespec's management believe that this change in operational emphasis will enable their engineering teams to focus squarely on verification instead of spending more time fooling themselves that architecture and design are first-order factors to overall performance and quality.
"Since the largest percentage of project resources is spent on software quality assurance (SQA), the team concluded that they weren't impacting the software development process by designing with so-called high-level languages," says Sathyam Pattanam, Bluespec's vice president of engineering. "The team just kept saying: we've got a verification problem, not a design problem, so where's the value?"
As principal engineer Ed Czeck notes, for example: "We've realized that using high-level programming techniques like strong typing only leads to one thing: type errors!"
Adds Shiv Tasker, chief executive officer of Bluespec: "Strategically, this is a sound decision. We get tremendous savings with offshore SQA engineers. Brilliant!"
Comments Rishiyur S. Nikhil, founder and chief technology officer of Bluespec: "I'm really enjoying spending 110% of my time on PERL scripts to enhance the new development environment. It beats thinking about technology strategy and the architecture of our toolset."
"I've longed for the good old days of the 8086," enthuses Elliot Mednick, principal engineer at Bluespec. "You are not a true engineer unless you can use overlays. I have always believed that you have to have complete control over argument passing, the stack protocol, and register assignment."
Joe Stoy, famous computer scientist, Bluespec founder and principal engineer expresses relief: "I'm relieved. To be honest, my long history advocating high-level languages and formal methods was solely a cynical gambit to obtain academic tenure when I was at Oxford. We can write assembly language – a common language we all share from training in school, as fast as any other language – the design time is exactly the same. Anyone can see that it's SQA's job to get more productive – they're the bottleneck!"