http://media.hardwareanalysis.com/articles/small/10580.gif" alt="Hyper-Threading Technology">
BAPCo's SYSMark2002 is a perennial, not to mention controversial, favorite. It generates a performance score based on scripted runs of several popular office and workstation applications, including Microsoft Office, Adobe Photoshop and Premiere, and other popular applications. One of the best features of SYSMark2002 is that it runs multiple applications at once, unlike its predecessors, so essentially it is much more realistic and representative of how an actual user would work behind the PC. As a result, we see the system with Hyper-Threading producing a slightly higher score. Granted, 3% isn't much to get excited about, but it's a start.
http://media.hardwareanalysis.com/articles/small/10590.gif" alt="Hyper-Threading Technology">
RC5 is a distributed computer project that attempts to search for a single 52-, 64-, or 72-bit key using brute-force computing methods. It's a fairly decent test of pure arithmetic throughput, and we see some interesting results here, especially when we consider that the RC5 client is multithreaded. Even with this added advantage, it is no faster whatsoever on the Hyper-Threading Pentium 4. This is due to the fact that, the FPU is being used up by one thread completely, with two physical CPUs, and two FPUs you should probably see a difference in performance.
http://media.hardwareanalysis.com/articles/small/10589.gif" alt="Hyper-Threading Technology">
ScienceMark is also a fairly demanding benchmark, and is not multi-threadeded. As a result, when we run it on both the Hyper-Threading and non-Hyper-Threading Pentium 4, we see no difference between them.
What we've tried to convey with the above three graphs is that Hyper-Threading can by no means be summarized generally. We've shown you that in some multi-threaded applications it can improve performance (the first graph), whereas in some multi-threaded applications it doesn't help at all (the second graph). We've also shown that if an application isn't multi-threaded, you shouldn't expect anything at all (third graph). So far it seems like Hyper-Threading isn't doing much. But we've yet to explore Hyper-Threading's greatest strength: multitasking.