Next week, Intel will host a Developer’s Forum during which we’ll find out a good deal more about some of their future products.
Here’s what you need to know:
Coppermine is only going to get up to the neighborhood of 1 GHz, maybe a little more.
The next performance chip from Intel will be Willamette. Willamette will be out in the fall/winter. Willamette will start at over 1Ghz and have a 200Mhz memory bus speed. Willamette will debut at .18 micron, then migrate to .13 micron sometime in 2001 and pick up copper interconnects at that time. Expect Willamette to eventually run up to 2Ghz or so.
Willamette will require you to rip out your motherboard and memory and possibly more. Although Willamette is supposed to be a new design; there are hints it’s not going to be a whole lot different than Coppermine: the improvements look to be more in the bus area. Don’t wait around for Willamette; it’s going to be well into 2001 before you’ll see the kind of system at the kind of price you’ll want.
Here’s a new word for your vocabulary: Tinma. Consider Tinma a Celeron II with built-in graphics. You probably don’t like that idea. You’re going to like the following even less: some of Intel’s mumblings on the subject seem to indicate that Tinma will eventually replace Celeron. No need to panic yet. We will still have Coppermine-128s. Tinma won’t debut for another six months.
It’s not clear Celeron will completely go away since AMD will have Ath-Celeron equivalents out there. But this could be bad news for gamers. Intel has hated the idea of Celerons cannibalizing the sales of its higher-priced processors, and Tinma could well be the way Intel really separates the product categories into high-end and low-end processors. It seems extremely foolish to me since such a strategy leaves the middle to AMD, but Intel hasn’t been too bright lately.
What does this mean for you?
We seem to have technologies piling on top of each other, without any one getting fully developed. Take the Athlon: we’re going to have three generations of motherboards for it within about a year. Intel looks like it’s going to debut a new generation of processors only a few months after getting out what will be the first real motherboards for the previous generation.*** It looks like first .13 micron will take place within eighteen months of first .18 micron.
We’ll be jumping from 100 to 133 to 200 to 266Mhz memory buses in the course of a year, with 400Mhz following fairly shortly thereafter. During all this, expect to see things like PCI standards changing, too. Over the next two years, if you want to keep up, you are going to have to junk the core (CPU, motherboard, memory, probably more) of your computer in one shot, piece by piece replacement won’t work.
This has to be a nightmare for manufacturing these products. How can you get back your development costs on something that only lasts six months? Maybe that’s why Intel hasn’t shifted everything over to .18 micron very quickly; perhaps some of these plants are being held back for .13 micron conversion. What does AMD do?
In a year, they will have gone through .25 micron, .18 micron, .18 micron with ondie cache, .18 micron with copper interconnects, getting a huge fab plant operational. Any rest? Nope. As soon as you have .18 micron Athlons getting cranked out, you have to start thinking about shrinking to .13 to compete with Willamette. It doesn’t look like AMD has a follow-up 32-bit chip to the Athlon; the next step for them is the 64-bit Sledgehammer, and that looks like a 2002 product.
Nonetheless, it looks like 2000 will be the year to upgrade those PII/Celeron systems, with Coppermine/Celeron II the first half of the year; on-die cache .18 micron Athlons the second half. After that, the next “value” upgrade doesn’t look like it will happen until late 2001/early 2002.
***We may have some more problems getting Solano2 motherboards. See
Real World Tech for details.