http://media.hardwareanalysis.com/articles/small/10584.gif" alt="Hyper-Threading Technology">
SPEC's SPECviewperf software is a high-end CAD/CAM/CAE benchmark that we make use of regularly as a video card benchmark, however it is also very CPU-intensive. It runs a total of six different tests; only one is shown here. Again, we see no difference between the Pentium 4 with or without Hyper-Threading when the application is run by itself, but when we attempt to run a second process at the same time (in this case, we used Windows Media Player to copy and encode music from an audio CD), the Hyper-Threaded CPU is able to share its execution resources such that the total time to complete both tasks is reduced noticeably.
http://media.hardwareanalysis.com/articles/small/10585.gif" alt="Hyper-Threading Technology">
However, Hyper-Threading isn't always going to make a difference, even when two distinct processes are running simultaneously. We encoded the same audio CD while running another of SPECviewperf's tests, and observed no noticeable difference whatsoever.
In summary, what we want you to take away from all these benchmarks is that, in fact, the performance of a system outfitted with a Hyper-Threading Pentium 4 cannot be simply summarized. Anyone who says it can isn't presenting the entire picture. There are simply too many variables, too many possible situations, applications, configurations, and so on. There are situations in which a Hyper-Threading Pentium 4 can achieve extraordinary performance, we measured up to +40%, and there are probably an equal number in which it will make no performance difference at all. If considering a system based on a Hyper-Threading Pentium 4, the wisest course of action is to explore your intended use, determine whether Hyper-Threading will actually make a difference in your specific case, and then decide if it's worth the costs.