If (pretend) I test the DEC CX on PI-120mhz (in DOS-MODE) and it took 2 cycle. Will it be the same if I test it on- Windows 98? Does Operating System affect opcode speed?The operating system can't change the speed of your processor (just imagine that ;)). So that certain instruction will always take 2 cycles on your processor.
What about RAM? does more ram speed up opcode's speed?The opcode will still take x cycles to complete, no matter how fast your RAM is. You might, however, experience some performance improvements when using faster RAM as opcodes have to be written to/read from RAM, but I think you'll hardly notice anything when just executing code. You might try Intel's VTune (you can grab an evaluation version at their site) for performance analysis of your programs (Windoze only).
NOT quite, my friend If OS is in 32 bit protected mode with pageing enabled (but with LDT/GDT also) it sometimes has to bring a new page on ram from storage or to change sone GDT selectors...then the instructions will take A LOT MORE to execute ... hopefully this will be only once ... :D but there is more there as one has cache lines to count for... If a read/write is out of the cache lines ... then a DELAY up to 32 cycles CAN APPEAR...more thing like this happen if you read the same cache line (in a burst) too soon... (i like that "too soon" and some docs say also: "rather one then zero" ) Things are getting worste and worster as time goes by as "modern" Micros try to mimic speed by CACHE tricks and to have more and more OS like features on chip...making them specialized for some tasks ONLY and SLOW them down to 285 on doing GENERAL things ;( Ipocrites, COMMERCE and COMPETITION my friends will KILL this planet...sooner or later... i just hope to get my share of babes before ;) This message was edited by BogdanOntanu, on 4/29/2001 8:39:02 PM
More RAM does not speed up an opcode. It reduces the amount of page swapping. That obviously means if your program doesn't swap any pages during its lifetime, adding RAM won't help.