>>>>> "KF" == Ken Fox <[EMAIL PROTECTED]> writes:
KF> I think this is why storage-to-storage architectures have lost KF> favor -- today's compilers are just too smart. Possibly with a KF> software VM the memory pressure argument favoring registers isn't KF> strong enough to offset the disadvantage of requiring smart KF> compilers. compilers are smart today because they have to be. risc designs dumped more optimizing work on them (instruction reordering, branch prediction, pipeline issues, etc.). S2S is just CISC. risc won out because CISC machines just did too much instruction and data loading for the amount of work done. the bus bandwidth was eaten by all the data address fetches and such. just look at a vax instruction and see how much bandwidth is needed to do one instruction. risc design is in effect exposing the microcode to the compiler and letting the compiler figure out how to optimize things. why do you think the 86 is so hard to build and make fast? it has to effectively convert a cisc instruction set to a risc internal engine. and compilers are hard fo rit since they have to know too much about the internal pipeline design. AMD and intel's designs are different and the compilers don't know how to handle that difference. and please don't bring in hardware comparisons again. a VM design cannot be compared in any way to a hardware design. the way compilers handle registers can be compared as that is related but hardware can do things in parallel and software has lower costs (as in silicon real estate). uri -- Uri Guttman --------- [EMAIL PROTECTED] ---------- http://www.sysarch.com SYStems ARCHitecture and Stem Development ------ http://www.stemsystems.com Search or Offer Perl Jobs -------------------------- http://jobs.perl.org
