I recently upgraded my partner's computer (as she had been complaining about how slow it was) from a 2.66GHz (512K L2) 512MB 533MHz Northwood System w/o HT, to a 3.40GHz (1MB L2) 512MB 800MHz Prescott System with
I did a format and clean install of all relevant software after the hardware upgrade (which required a new motherboard to run the Pressy) and when I had finished, I ran a couple of benchmarks to determine how much more performance the upgrade had produced VS the old set up. In addition, I also timed the boot-up speed of the OS, from the moment the computer was turned on, to when the last taskbar icon (Norton Firewall) had loaded.
Here are my results:
PCMark04 Score: 3952
Boot-Up Time: 146 Seconds (2:26)
Now, compared to many PC's, these scores are mediocre at best, but compared to the previous scores I got from this PC when it had the 2.66GHz Northwood set-up running, these scores were a fairly big improvement.
Now, here comes the part about HT (hyperthreading) and why you should turn it off if you (like my partner) run Windows 2000 Pro on your PC...
Disappointed with the boot-up speed of my partner's PC still, I looked around the internet for some Win2k tweaks that would speed things up and during my search, I came across an article that talked about Win2Ks poor management of Intel's HT. Apparently, Win2K doesn't recognise HT and only sees an HT enabled CPU as two actual, physical CPU's, which clearly they're not. This leads Win2K to use HT in a highly inefficient way, that, in some circumstances can lead to HT actually decreasing system performance, rather than increasing it. WinXP and various other Windows versions post Win2K can "see" HT for what it is and use it properly, but Win2K it's self doesn't do this.
Having read this information, I then decided to disable HT in BIOS and then re-benchmark the PC to see if turning HT off made any difference. Here are my results:
PCMark04 Score: 4193
Boot-Up Time: 142 Seconds (2:22)
The results were clear... A 241 point (6.1%) increase in PCMark04 Score. A 430 point (34.2%) increase in PC Pitstop
Score and a boot-up time saving of 4 seconds (2.7%).
Now, these are not massive performance gains as I'm sure you know, but they were the sole result of simply disabling HT and why I think that if you run Win2K and have a CPU with HT that is turned on, you should, at the very least, do some benchmarking with it on and with it off and see for yourself which is better.
You may have already known about this, but if you didn't, it might help you to attain the best performance from you Windows 2000 Pro, Intel HT system.
Any comments welcome