Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
Prices: Mainboards ABIT ASUS Chaintech Shuttle Soyo Tyan CPU Intel P4 2.4C-800 P4 2.6C-800 P4 2.8C-800 P4 3.0-800 P4 3.2-800 AMD AthlonXP XP 1700+ XP 2000+ XP 2400+ XP 2500+ XP 2700+ XP 3000+ XP 3200+ Athlon64 Athlon64 3200+ Athlon64 FX-51 Opteron Opteron 240 Opteron 242 Opteron 244 Opteron 246 Memory Corsair Crucial Kingston Mushkin OCZ Search Prices:

The big issue on everybody's mind when HT was released on the Xeon was that there were benefits and there were drawbacks in non-optimized applications and the drawbacks outweighed the benefits at least in the typical desktop applications. Other concerns were regarding licensing fees for multiple processors and in general, there were wild speculations about HT being potentially able of carrying out totally different instructions within one clock cycle courtesy of the two logical processors.

The veil has lifted by now and even though the fog still lingers, there is enough information on technical issues, there is a ton of benchmarketing available as well and in the end, whoever can read, will be able to make up his or her mind about the sense or nonsense of HyperThreading. From a design standpoint, HT is a very elegant solution to increase processor performance by opening up the "front end" of the processor, that is, by enabling different threads to be funneled into the execution units like a zipper would do. Now, even though I like the example of a zipper, looking at the stats within the screenshots of the Windows Task Manager, it is clear that current OS are not restricted to two threads, as a matter of fact, the average we saw were several hundreds of threads that were running simultaneously and that is a pretty big zipper.

Suffice it to say that there are applications that will see performance hits in the order of up to 10%. Suffice it to say further that the majority of applications that are not HT optimized will be penalized by about 1%. By the end of the day, however, we will find that the specific applications that are historically really time consuming and / or have taken up enough system resources to effectively cripple any PC will not only run in the background but also see a performance benefit from HT that can be up to 30%.

There is no point in trying to run Quake3 faster through HT. In gaming applications, the bottlenecks are at the level of the graphics cards, in Winstones, they are at the level of the I/O interface. Then, why in the world would anybody even bother? I think, we did answer this question and if there are still doubts, think of the power of cheese. Aside from that, we will still need some software optimizations but keep in mind that if the code gets too good, there will be no more benefit from HyperThreading.

next page:    => more =>