HD is much, much slower than RAM: we're talking nanoseconds versus miliseconds! And you don't accidentally miss the 'bus' so to speak: you will have to find a file first on your HD (it's physical position) then get the reading equipment adjusted to read from the particular 'lines'; and finally you need to wait until the beginning is 'under' the 'reading tip'. With good RAID systems the problem is somewhat less (stuff can be read from multiple tips at the same time; from phyiscally different HD's --> in fact RAID means duplicate HD to increase read/write speed; but mostly to ensure persistency in case of hardware failure).
And actually CPU's also use another, yet even faster type of memory than RAM --> TLB (Translation Lookup Buffer: used to translate page/segment and similar addresses into phyiscal RAM addresses; which in turn serves as a safety layer as well as help memory management (optimize for frequently used pages/segments)...) and processor cache. But that stuff is so insanely expensive you can't make decent amounts of RAM from it and expect to find enough buyers to make a profit. The key is much more parallel look up and read/write operations; as well as the associative 'behaviour' of the electronics.
Bookmarks