|
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 |
LOSTCIRCUITS |
|
| Pentium4 3.06 GHz GHz HyperThreading and the Non-Parallel Universe | |
| (Review by MS, Nov. 18, 2002) |
Test Configuration
|
|
|
| ||
Intel D850EMV2
i850 Dual Rambus chipset. HT can be enabled / disabled in the BIOS.
Click image for larger view and more details
A Brief Word About Benchmarking
Benchmarking of HT is plagued by a number of vagaries, in the optimal scenario, both applications running simultaneously should have the same runtime to avoid differential weighting of the individual scores in the overall outcome. In reality, the situation is not quite that bad. An analogy is the 3rd grade equation where two cars are approaching each other. Starting distance is 50 miles, one car (a) can run 50 mph, the other one (b) only 10 mph and if they are running at the same time, they will only achieve 1/2 of their max performance each. Moreover, to be fair we'll have to impose a limitation of 25 miles distance for each car. Anybody can now do the math and figure out that it does not matter whether car (a) or car (b) starts first or whether they both start at the same time. The run time will always be 3 hours. If the total distance is changed to 40 miles, we'll have to leave the max distance for car (a) but have to reduce the distance for car (b) to 15 miles. If we don't do that, we run into some real problems.
The example at hand elucidates another problem, that is, which is more important, the speed at which both cars are traveling or the overall time until they meet? It is a nice exercise to do all the calculations, good luck!
In reality, devising two different applications with identical runtime while either enabling or disabling HT borders on the impossible but we can still get some decent approximations.
next page: => Performance =>
