Life has been very quiet in this neck of the woods the last month or so.
It’s going to be that way longer in the overclocking world. A lot longer.
The overclocking world works a bit differently than the rest of the computing world. We’re not particularly interested in new products, we’re interested in cheap newish products.
And since we push the envelope, developments that might not even be noticed or just be annoying to the average computer user are critical to us.
The next year or so is going to see a lot of developments, but they won’t be significant and/or relevant developments to the overclocking world. This will be either due to a high price, and/or a modest increase in performance for the expense.
Some items won’t deliver much, period. Others will become interesting, but only after the price drops quite a bit. Yet others will eventually become important, but not in their initial iterations.
This will occur among a group of people that are increasingly reluctant to spend a lot and get only a little for it.
Perhaps more importantly, the outside world will have more and more to say about how this cyberworld conducts itself.
We’ll go through the next year, and tell you what you’re going to see, may see, and not see.
We’ll see a drumroll for Athlon64 as summer ends and fall begins.
We’ve talked about this many times over the past couple months. To summarize:
From what we’ve seen from Opteron overclockability, AMD still doesn’t have the SOI process down pat, at least not at 130nm. It’s going to be hard to get chips running much past 2GHz. We’ll probably see a 2.2GHz chips, but odds are those will be cherry-picked processors.
You may say “Well, that’s roughly about what Athlons can do at default voltage,” and you’d be right, but it doesn’t look like AMD will be getting any big performance benefit for this generation.
AMD says the $46 million they paid IBM to help them out with SOI went to 90nm development. Based on IBM’s promises to deliver a 3GHz G5 to Apple in a year, there is reason to be hopeful about AMD’s 90nm prospects.
But that’s not what’s going on sale in September.
The PR ratings for the Athlon64 look inflated in 32-bit mode. To justify the ratings, the Athlon64 is going to need to run in 64-bit, and, sorry, Linux fans, for the desktop, that means Windows x86-64.
However, running in 64-bit doesn’t mean just using an x86-64 OS; it means x86-64 applications and games, too. Given the relatively weak performance of Hammers in 32-bit and AMD’s less than enviable track record in getting consumer software written for it (remember 3DNow?); this will be AMD’s Achilles heel, and don’t bet Intel isn’t slashing at it behind the scenes.
Over the next year, it will probably be a better use of one’s time to keep track of the availability of x86-64 software rather than the performance of x86-64 hardware.
Price/performance will be a huge factor. It’s hard to see how the initial generation of Athlon64s are going to be significantly better than current Athlon systems. From earlier surveys we’ve done, the vast majority of Athlon owners want to see at least the equivalent of 4GHz before they’ll start thinking about changing platforms, and that will be very hard to do at 130nm.
Despite this, it looks like AMD intends to charge premium prices for these chips. Given the . . . err . . . frugalness . . . of the typical AMD buyer, this is not a good combo.
Adding to the difficulty is that early buyers will get short-hopped on the motherboard platform. Next fall’s platform will be a DDR/AGP motherboard. Next summer’s will be a DDR2/PCI-Express motherboard. This reality isn’t going to encourage those who are already thinking twice about changing at all.
Provided AMD doesn’t suffer any delays in transitioning to 90nm process technology, the combo of 90nm Hammer technology and the new mobo platform should occur next June. By that point, we should have a pretty good idea how well x86-64 will fare.
It’s not a terribly promising picture for AMD. However, matters don’t look too hot on the Intel side, either . . . .
Fe-Fi-Fo—Oh-Fu . . . .
The reports on Prescott processors chewing up 10-15% more power than expected is one of those things that affect overclockers more than others.
When Intel is talking about 90-100 watts at default, we’re looking at 130-140 watts in a typical overclocker setup. If this doesn’t concern you, it damn well ought to. For power per cubic centimeter, we’re going to see numbers that make the AMD Thunderbirds look cold.
A little known fact about the Pentium IV is that a disproportionate amount of the power it draws, well, leaks. This is a big reason why Intel is so interested in new technologies like strained silicon.
Well, whatever they’ve done with Prescott to try to reduce the problem isn’t working so far. The big problem for Intel isn’t the possibility that some current Intel mobo users may not be able to use Prescotts with their boards. It’s that Intel has lost control of its process technology: new chips are heating up sooner and faster than they had expected.
If they can’t tame this beast, and quickly, it will probably have an adverse affect across the board. It could mean Intel will have to start playing the delay game, and reap bad publicity if the new processors have a habit of overheating or just end up needing too noisy cooling. Remember how big a negative it was for AMD to have its processors called furnaces.
For overclockers, and especially Intel overclockers, who are as likely as not to stick with the retail fan, these heat levels will become bottlenecks.
Something you ought to keep in mind is that Prescott is a two-way transitional chip. It’s only supposed to ramp from 3.4-3.8GHz in socket 478, then hang around for a short time as a socket 775 chip until Tejas shows up. for socket 478 was never meant to have a long-life, anyway. It’s a transitional
When they do come out, these Prescotts will cost a lot, as you would expect. Even if Intel gets everything under control, and quickly, no Prescott will become affordable until the beginning (if Intel releases lower-speed OEN Prescotts) or spring (if they don’t) of 2004.
So even if the problems above all get fixed, you won’t get a crack at Prescott for six-nine months, anyway, and if there are problems, you may not want to take a crack.
There’s another problem which will affect both AMD and Intel. The CPU isn’t the only place where changes are coming.
Newer But Not Better
The new mobos coming out a year from now will have two critical differences from the mobos available today.
First, they’ll use DDR2 rather than DDR. Second, PCI-Express will be introduced as a new video standard.
So late next spring, just when the CPUs reach the right price and/or performance level, the world will have to break in a new memory AND video standard at the same time.
Actually, there will be 2 1/2 new standards, SATA should be in full force by then, but it is likely you won’t be required to replace your current HDs like you would with memory or video.
This is likely to cause more than a few problems.
Initial extortionate pricing should be taken as a given.
Technical problems should also be expected. For instance, Windows won’t innately support PCI-X until “Longhorn” comes out in late 2004/early 2005, and kludges are already being contemplated.
As is usually the case with the intial iterations of new technology, neither DDR2 nor PCI-X video is likely to be much better than the last versions of what it’s replacing, and it will probably take the second generation of products to make a real difference.
So when you’re starved for something new, you’ll get to play guinea pig.
As if life didn’t suck enough, outside forces are out to cramp your cage.
Attacking The Magic Kingdom
Computers are tools, not statues. You don’t buy them to admire it (well, at least not most of the time), you buy them to use them.
And let’s face it, many people bought computers in the first place for all the free stuff they could get online.
Forget the nasty lawsuits and the even nastier threatened legislation. The P2Ping environment is a lot more hostile and less productive than it was even a year ago, with loads of decoy files and more and more people leeching rather than “sharing.”
This will only further deteriorate in the year to come.
RIAA and Company will continue to vigorously apply both carrot and stick. It will provide carrots like BuyMusic.com, which is essentially a Windows knockoff of iTunes. It will continue to pursue the lawsuits and fine-tune their legal arguments and settling-out-of-court strategies.
Expect the movie people to get a lot more lively as DVD-recorders reach the point of mass-acceptance.
Expect both to put on the full-court press in Congress and say, “You gotta do something.”
And that “something” will be DRM. Next year, there will be a lot of talk about TCPA and Palladium.
What the content providers will want is simple: a U.S. mandate to have all computer equipment sold after a certain date to include DRM technology.
RIAA and Company have very carefully laid out the ground for this. They’re going to say, “Look, we listened to the legitimate complaints. Look at all the songs we’re now offering digitally, but people are still doing this. We’re even suing our own customers, but that isn’t even helping a lot. You have millions of people out there with no respect for the law. Now what do you want to do? We can’t sue them all. Do you even want to try to arrest them all? We have the right to demand protection under the law, so what are you going to do about it?”
And DRM is likely to be the answer Congress comes up with. It’s the easiest solution from their perspective; keep people from breaking the law in the first place.
Prescott will be DRM-enabled. If AMD hasn’t already done the same thing, they’ll certainly be ready to do so. The other equipment makers may well fight such a mandate becoming law, but they’ll certainly comply; they aren’t going to lose 40% of the world computer market over this.
Could it become a world standard? Pretty good chance of that. The European Union isn’t going to raise much fuss over it.
Will there be a politically significant movement against this? No. There’s no organization worthy of the name out there, and even if one formed, what they want is outside the political norm.
It’s not often that you find an issue liberals like Maxine Waters and conservatives like Orrin Hatch agree on, but this is it.
A hypernova of hysteria will reign in geekdom, but no one else will care. The hardcore will go underground. The older ones will start buying their music digitally. The younger ones will copy each other’s older CDs (if possible), or something else to do.
Don’t be surprised if somebody figures out a way to put an FM tuner in a PC and “tape” CDs off the radio.
There will be surprise developments and twist and turns, but the eventual outcome is inevitable. What’s going on now will be criminalized, and because we don’t want that many criminals around, people will be stopped ahead of time.
This will put a chill on computer upgrades for those who bought the thing in the first place primarily for that. The industry will survive. Maybe not too good for those making recorders, but they’ll find something else to do.
You just can’t expect a secure future when it relies on theft.
Five or ten years ago, any or even all of these events may not have mattered very much to the computer industry. Back then, with a few exceptions, there wasn’t the general feeling of “enough,” and even when it was around a bit, it went away soon enough as new equipment made a real-life difference.
That was then, this is now.
Probably the biggest change in recent times is the spread of “enough” even around here. When even the loons don’t want to go crazy, you have a problem.
The next year to eighteen months will give us products with lots of new jargon and buzzwords, but they really won’t be much better than what we have now.
In all honesty, you could go away for a year, come back, and really not miss anything real-life substantial. You might even push that to close to eighteen months.
Now that’s a drought.
Is this another one of “Ed is always telling you to buy later.” No. Ed is telling you to buy now (or fairly soon) if you’re behind the curve. Whether you prefer AMD or Intel, you can update yourself with ripe, mature products that will hold you rather nicely through this long, troublesome dry spell, and save you a lot of headaches.
Then in late 2004, early 2005, you’ll be looking at a whole new world that’s finally better enough than today’s for you to buy into.
That’s good advice for you, but it’s hardly great news for those of us covering this.
Those websites that just review products are really going to have problems. They’re going to have to make much ado about nothing for a long, long time.
Yes, there will be news, and products, and no doubt an increasingly desperate marketeering tone when talking about them, but it will be preached to an ever-increasingly skeptical audience saying, “Why should I buy this?”
And they won’t have an adequate answer.
It will be interesting.