I’m starting to get a lot of mail from people who are getting itchy. The CPU prices for AMD have reached desirable levels, and folks are reacting to that in a way
that makes salmon during mating season look ambivalent.
“Old” AMD Durons and TBirds up for sale. First generation boards still rule the roost, which means 100Mhz FSB and native ATA66 support (without add-on circuitry).
A7V appears to be the only board with a reasonable degree of stability for most people. Second generation boards have ATA100 support (if you do some digging), unimpressive so far.
Within a month:
KT133A motherboards start coming out. Will run at 133Mhz FSB for real. Looks like 133Mhz FSB will get you 3-5% improvement over KT133 boards. Let’s you use PC133 RAM. No idea how stable these things are in real life.
Over the next three months:
DDR boards and memory become readily available. Initially, AMD76x and Ali Magik1 boards available. AMD76x boards appear to plod along, with not much improvement over SDR systems. Ali results appear to be all over the map.
Towards the end of this period, Via DDR boards should become available. Stability even a bigger question mark. DDR memory itself scanty, very high prices; probably will decline quite a bit Feburary-March.
Maybe over the next three months:
AMD very quiet about Palomino, had promised at least an intro by the end of the yeat, no signs of it yet. If Palomino is dependent of isotopically pure silicon (which I doubt, at least for the
initial ones), there’s been delays in making that.
The Palomino Puzzle
We don’t know two items we need to know in order to assess the overclockability of these chips:
- Whether or not the multiplier can be adjusted
- The FSB of these chips
If you had to put a gun to my head right now, I’d say the multiplier probably can be adjusted, and the chips will be set for a 133Mhz FSB.
The 12.5X barrier
If the Palominos are all 133Mhz FSB CPUs; you’ll have two ways to overclock them: increase the FSB or increase the multiplier. Initially, you’ll probably be able to get to 145-150Mhz/290-300Mhz with good DDR RAM, probably about the same or a little better with PC133.
However, you need a motherboard that can run at those FSB speeds to even think about overclocking that way, and even if you just do a multiplier change, you still need to run at 133Mhz. The motherboards (whether they be 133A or DDR) that can do this aren’t out there yet.
You might say, “Well, can’t I just increase the multiplier on my current board?” Sure, you could, but what’s the maximum multiplier you have? 12.5X. What’s 12.5X100? Not 1.5Ghz or better, which is what you would have with a mobo that could handle 133Mhz FSB.
Might the multiplier range be extended with a BIOS update? Maybe, maybe not.
The AMD datasheets are rather mysterious about this. They say more than 12.5 is achievable, but they say the signal for it is the same as for 12.5. That tells me something within the CPU has to trigger more than a 12.5X multiplier.
I could be wrong on this, but if I’m not, you may find yourself having to buy a new motherboard in six months to go along with that Palomino upgrade because you couldn’t wait a few weeks now. If that doesn’t bother you, no problem. If it does, then you should wait.
A few days ago, I wrote an article that was very skeptical about buying the 133A immediately. I still am.
If you haven’t bought yet, though, and the idea of upgrading to a Palomino later on is important to you, the 133A may have a role to play.
The KT133 is not effectively upgradable to a Palomino
The key word here is “effectively.” Yes, it can be done, but to get the full benefit from this chip, you need at a minimum 133Mhz FSB speed, and the KT133 can’t do that.
If upgradability six months down the road is no big deal, then don’t worry about this. If buying a new mobo in six months time is no big deal, don’t worry about this. If either is a big deal, worry about this.:)
I Still Prefer DDR But . . .
At the very least, let’s wait and see how well these KT133As work. They should be coming out in the next few weeks. We’ll test and work with them. If we find a good one, we’ll let you know.
We’ll also be working with DDR boards, we find a good one, we’ll let you know that, too.
Once there are good 133A and DDR boards available, then you’ll have a decision to make between the two, and we’ll talk about that then, too. You’ll have to decide then whether the performance improvement of DDR is worth it to you.
But for right now, we don’t know when or what we’re going to say in a couple months. There’s too many things we don’t know right now.
The only thing we know for sure is that you’re buying old technology that’s going to be obsoleted shortly, and increase your next upgrade bill if you buy now.
“But 1Ghz chips are SOOOOOO cheap right now!”
Yeah, and they’re going to get cheaper, too. Look at the spectrum of TBirds available today; the low-end is under $100. No matter what happens with the Palominos, you’ll be able to fall back on these chips later on, they’ll run fine at 133Mhz.
But unless you just like buying new motherboards every few months, why not show a little patience and get something you can get at least a CPU upgrade out of?
“I Need A Speed Rush”
Look, if you got a P100 or 200, you are hurting. For you, there’s probably more pain than gain in waiting.
But most of you aren’t exactly in that position. If you’re at around 500Mhz or better, don’t expect a big “whoosh” by hitting 1Ghz from sheer perception. You aren’t going to get it when you turn the machine on, or for most things.
Human being are actually pretty bad at determining computer speed; it takes a big difference for it to even be noticeable.
Recently, I’ve worked with a Celeron 800. Right now, I’m using a TBird at 1.1Ghz. For the work I do, I can’t tell the difference between the two, and outside of maybe a game, I doubt you could, either.
Let’s face it, for most of you, this is a discretionary purchase. You can wait if you want to, certainly a month or two at worst, maybe just a few weeks at best.
“There will always be something to wait for down the road”
There’s a major problem with that sort of thinking; there’s a huge difference if that “something” is a few steps or a few miles down the road.
The pace of technological improvement is not steady. It tends to make big jumps quickly, then slow down or even stop for a while. The smartest strategy is to buy after a big jump, not before. You buy now, you’re buying before.
Look at the Coppermines. During the first half of 1999, we got pretty rapid improvement, which then slowed down. We expect the same to happen with the Palomino chips, big jump first half of the year, slow improvement second half.
After that, we expect the next big jump to be .13 micron Willys/Clawhammers sometime in 2002.
We think you can get to 1.5Ghz affordably sometime the first half of the year; then the pace will slow until 2002.
Is The Problem Your Computer Or You?
This society is built on “buy, buy, buy.” I get so many emails that can’t be described other than folks in a buying frenzy; money is burning in the pocket.
We see our job as giving you the information you need to make informed purchases. You certainly can buy a system now for sane reasons. We just think many of you would be happier buying a bit later, but you make that decision. We just want you to know what’s coming so you won’t be blindsided by some technological change, and have regrets later.
If you buy a computer every three years, it probably doesn’t matter very much.
However, if your equipment is less than a year old, you really should ask yourself what the rush is. If the answer is, “I need a rush,” that’s not a good answer, especially when you’re likely to crash afterwards when you get disappointed.
The issue here is not whether you should buy or not, just when. If you gotta get a rush, at least spend your money on things that will get you one. Go get cable modem or DSL, that will get you a lot more “oomph” than any hardware improvement.
Patience is good. It gives you the chance to get more satisfaction for your cash. I’m not against you spending money, just against you spending it a little too soon.
If you’re saying to yourself, “I have to have 1Ghz now” ask yourself how you’ll feel in three months when people are saying, “I have 1.5Ghz now.”
Many, if not most, of you buying will find good reason to buy now. That’s fine, whatever works for you.