Jim,<<<P4 unveiled... >>>
I'm not an engineer, but more likely I have more $ invested than Darek Mihicka, the writer who for some really unimaginable reason is trying to wreck my investment.
I think I understand what he is trying to say in his lengthy psuedo technical diatribe:
1. Even though the P4 is capable of executing 6 billion integer arthimetic operations per second, clock speed is not everything.
Most compiler tools are written for older processors and if the compiler tools are not rewritten to the exact recommendations of P4 designers, programs will most likely execute much faster on older processors.
2. The Athlon chip is designed to be far more accommodating to legacy code. In some circumstances, it does not force programmers to code in a certain way. They could continue to code (legally) in the way they are used to and those programs will run more efficiently under Athlon than P4.
3. Most Windows applications use lousy compiler tools to begin with, and when those compilers are not rewritten and take into consideration, fatal flaws in the P4, there could be trouble.
As an example, the P4 will encounter partial register stall, IF THE PROGRAMMER DOES NOT CLEAR THE REGISTER - USING THE CLEAR REGISTER INSTRUCTION !!!!!!
ATHLON ENGINNERS USE NEAT TRICKS SO THAT PROGRAMMERS DO NOT HAVE TO CLEAR THE REGISTER. (I assume there has to be some very small, negligble overhead to achieve this).
IN CONCLUSION THE P4 DOES HORRIFIC JOB OF EXECUTING LEGACY CODE. MUCH OF THE LEGACY COMPILER TOOLS (MOST PUT OUT BY MICROSOFT) ARE LOUSY TO BEGIN WITH - AND IF NOT REWRITTEN AND OPTIMIZED USING (POSSIBLY) ASSEMBLY LANGUAGE THE REULTS COULD BE THAT PROGRAMS WILL RUN MORE EFFICIENTLY ON CHEAPER PROCESSORS AND THEORETIC CLOCK SPEEDS WILL NOT BE REACHED.
What do you think, Jim, that at the next Intel's stockholders meeting, we suggest that Intel replace Craig Barrett as CEO with Darek Mihocka? This guy is brilliant. How do I know? Are you stupid, or what? This guy took great pains to tell us this.
You buy it, right?
Mary |
