There are plenty of hyperthreading articles out there, but you don’t need to read them, yet.
All you have to do is ask yourself, “Is hyperthreading worth an extra $500?”
For 99.9% of you, the answer is no. End of story.
“But I Can Look, I Can Dream”
And you can be a sucker, too. This is why:
If Intel can make its plan stick, only 3.06 or better CPUs will have hyperthreading. Looking at the roadmap, the 3.06 doesn’t look like it will become affordable (i.e. <$250) until Prescott is close to debuting, if it hasn't already arrived.
Prescott will have a more advanced hyperthreading system, and having a 1Mb cache certainly won’t hurt it.
This gives the 3.06 the dubious distinction of never being a good buy. Buy it now, and you’ll pay far too much. Buy it later, and you get near-instant obsolescence.
Hyperthreading is a nice “gimme” technology, and little more. You certainly shouldn’t pay an arm and a leg for it.
Spending $650 now or even $400 next spring for a processor just to get the first generation of hyperthreading is practically throwing your money away.
A Mismatched Technology
When you think about it a bit, hyperthreading is really more suitable for slower rather than faster processors. It’s when you don’t have the clock cycles that you need a more efficient processor.
Indeed, hyperthreading in the lower echelons of Intel processors would make them much more competitive against Athlons, and often give them the edge in a typical Joe Sixpack or Suit system.
Multiple Tasks Still Whack Performance
Take a look at this page.
What you’ll see are a number of activities run first as a single process, then run with a significant background activity.
Don’t focus on how hyperthreading affects the score. Focus on how much the scores drop, hyperthreading or not, when you add that second task. Hyperthreading just lessens the dropoff a bit.
If you’re hellbent on performance for a certain task, you don’t need hyperthreading. You need to just run that task by itself.
This Is Marketing Magic
I’m not against hyperthreading as a technology as all. I am against hyperthreading as a justification for a high price tag, and this is precisely what Intel is doing with it.
The technology is a modest (though very clever) improvement which actually delivers a minor (but relatively large) bang for a small silicon buck.
What it is being marketed as is the Big Bang which deserves many hundreds of your paper or plastic bucks. And it isn’t.
So ignore it until it comes in at a more appropriate price, which will come with Prescott, if not sooner.
But The 3.06 Runs Faster!
Some of the initial overclocking attempts in those few reviews that do that sort of thing anymore are frankly impressive, with results of 3.7GHz with high-end air.
Again, though, $650. Add a Prometeia, and we’re talking $1,000 for maybe 4GHz.
It would be nice to cut that figure down to size.
The Post-C1 Stepping
My suspicion on these chips had been that they represent the first “multiple VID” PIVs. Just so happens that Intel updated its Processor Specification Update today, and, in fact, that’s just the case.
Look at the bottom of page 8 and go on to the first half of page 9. See all those processors with entries “5, 16” in notes? They are all this post-C1 stepping, and I’d bet dollars to doughnuts that when they come out, they’ll do just about as well as today’s 3.06 for a lot less money.
When? At earliest, the end of January, according to this.
Is waiting a few months worth $500? Is an extra 400 or 500MHz worth $500?
For Those Who Say “Yes”
I know people want status symbols. I also am seeing more people as of late trying to buy themselves overclocker status.
I find a lot of that silly because only the shallow think that things make you more important than others or not. They don’t.
CPU status is even sillier because any “status” you get for your money is even more fleeting than most other purchases. The value of what you bought drops, fast. Today’s $650 processor is next year’s $100 processor.
Just think about it.
What About Other CPUs? Do They Have Hyperthreading, Too?
I’ve gotten a lot of contradictory evidence on this one. Won’t go into it now, but I suspect that turning hyperthreading on, at least for the newer PIV processors, is doable.
Take a look at this comment.
“What has been hidden from the user up until now is that Intel already integrated its Hyper-Threading technology into the Willamette core two years ago. Software developers were able to access this function right from the start for research purposes. It is not possible to activate HT on older P4 CPUs, because certain settings in these chips are permanently disabled.”
Hmmmm. How can software developers access a permanently disabled function?
There have been a couple people who apparently have somehow “turned it on” by accident for a moment, which is something you don’t find happening with the multiplier lock.
It may well require something like WPCREDIT, but I wouldn’t give up hope on this.
And of course, the second you can enable HT on a PIV that costs less than $200 (preferably $120), I start liking HT very very much. 🙂
Again, love the technology, hate the unnecessary price tag.