mystikmind2000: Funny how so often old games run more poorly on a modern PC than the 20x slower machine you first ran the game on 10 years ago! And even when the games performance does improve on a modern machine, it is like about 2% improvement when the modern computer is like 200% more powerful!
rtcvb32: Part of it's the chipset. CPU's used to run with hardcoded opcodes and the chips were bigger, now it's all microcode within the CPU. Also games back in the day all ran on 1 processor, and having 4x the processors does not translate into 4x the processing power, it's more like 2x with semaphores and all the permissions interacting to get access to the same resources.
Then there's the individual instructions. There were lists of how many cycles individual instructions used, going down from 10, to 8, to 5 meant not only when you bought your CPU how many cycles it did was faster, but the time spent on each instruction was faster overall due to optimization.
I'm sorta reminded of an old sorting algorithm [s]quicksort[/s]
ShellSort. Richard Buckland explained it basically being bubblesort, but going about the job considerably better. However it's not used today because caching is horrible for how far the indexes are to eachother, as well as it can't be worked on in parallel easily, once multiple cores/CPU's are around it was no longer an option, not because it won't work, but because it ultimately performed worse.
It is kindof like trying to make a horse run faster by strapping a car engine to its back (although tying firecrackers to a cats tail, 'does' make a cat run considerably faster)