| ??? 11/03/05 02:17 Read: times |
#103258 - Compiler improvements Responding to: ???'s previous message |
Charles Cohen said:
You're going to make a bunch of students with their heads deep in books titled "Advanced Compiler Technology" very disappointed... Wouldn't it be sad if we had to trust all future performance gains on the h/w designers, and think that the capabilites of s/w have stayed the same for 25 years? The point is that the Keil compiler for the 8051 architecture is a very mature product, as is the field of compiler design. That's not to say that there can't be any more improvements in these areas but earth-shattering, massive improvements in optimizations are going to be less common. It's just the law of diminishing returns. It might be easy to reduce a 1000-cycle routine to 500-cycles, but reducing 500-cycles another 50% to 250-cycles is much more than twice as hard. I think most optimizations that one can expect from a mature compiler are going to be very architecture-specific. So unless those students are reading "Advanced Compiler Technology for the 8051" I'd be surprised if they were able to improve upon Keil. Regards, Craig Steiner |
