If you looked at the actual machine code all those many years ago when I was doing some piddly programming in C+, until it was optimized it was full of dead space filled with no-op opcodes (do nothing, skip to the next opcode) since the programming language set aside large amounts of data space for all variables.
Microsoft programs in those days were notorious for size for this reason, very poorly optimized, and as a result dead slow since the processor spent most of it's time loading no-ops. Almost as stupid as the sequence so often found of "mask interupts" followed by "halt processor" that none of their programming languages filtered out. That sequence turns off the mouse and keyboard and stops the processor, only recourse is to power cycle to reboot. The theory was, like "reduced instruction set" processors, that it used fewer clock cycles to execute an endless string of no-ops than it took to write decent code that ran far fewer instructions. Sloppy programming at the very best, digital malpractice is more like it. I never did understand how using programming space to write the complex operations of CISC processors was any faster by using a RISC system, unless the CISC chip was from Intel and took half an hour to change stacks or something. Off-loading the complex operations onto the programmer is going to cause more trouble (and errors) than complex instructions executed in hardware on the chip, especially task switching. I'm not sure Microsoft ever did figure that one out, they had to buy Unix instead. _______________________________________ http://www.okiebenz.com To search list archives http://www.okiebenz.com/archive/ To Unsubscribe or change delivery options go to: http://mail.okiebenz.com/mailman/listinfo/mercedes_okiebenz.com
