Jeff Atwood had some ramblings a couple months back about the diminishing returns of inserting more and more cores into desktop CPUs. It is true that applications not written with a multi-threaded model will not be gaining any performance gain in itself other than having the OS push competing processes to other cores to gain a true multi-tasking environment. In fact, the slack afforded to allow one wayward process to "hijack" a core by spinning 100% in an endless loop is going to prove very useful - programs in the future are going to be just as buggy as they are now and in the past - other cores should still be able to carry on like a bomber with an engine shot out.
However, such is marketing hype that we tend to narrow down on the current "component of the year" and lose awareness of the system at large. One commenter, Gustavo Duarte, wisely points out my sentiments:
There is of course the issue of how much of a bottleneck the CPU is in the first place.
Disk I/O, RAM I/O, and various latencies are more and more the real bottleneck. Especially for desktop systems, where we often have 1 process running and 30 others sleeping, without much chance of parallelizing work.
Yes, there are applications that truly demand raw processing horsepower, let us not deny that. But the biggest problem with desktop systems, past and present and fhe future to come, has always been all the other slower components that make up the desktop computer. In my lifetime of computer usage, the moments when the system slows due to extremely CPU utilization versus the CPU waiting for the disk to load the necessary data into memory (or swapping out) is an very, very low ratio.
More often than not, like a kungfu master tapping his toes waiting for his disciples to get prepared and smacking them down with a single blow when they do come, the CPU is not doing much other than to wait for the necessary instructions and data to come.
This is most pronounced by the component with moving parts: the hard disk. Today's disks are getting faster and faster indeed. But compared to processors, they are turtles and snails. The bulk of system slowdowns are easily associated with the hard disk LED blazing away, as they struggle like an underpaid coolie carrying bags of rice by order of the supervisor (CPU) who does not move as much. This only gets horribly disproportionate as the inverse-pyramid pattern gets wider: more fast components served by less slow components.
Just how many people will buy quad or oct core CPUs in the future? And how many out of those will buy more than one hard disk? The tech saavy among us may know this and wisely purchase one or two additional disks to maximise disk I/O, but the layman consumer is unlikely to bother with that, and simply choose the biggest TB (terabyte) disk the vendor has to offer and break it into multiple volume partitions. Besides that, until disk sizes continue to shrink and fuel cells make their way into laptops for energy endurance, we should not expect laptops to generally feature two internal disks by default.
Will we see some kind of duo spindle disks or self-contained RAID technology emerging? Since I do not follow the hardware landscape as actively now, I don't know at this point. But I know the disk industry has alot of catching up to do.