As reported around the web, VIA is about to launch its latest Athlon chipset, dubbed the KT400, sometime in April. As the name implies, the chipset will use DDR400 DDR SDRAM.
This is ridiculous.
Why Is This Bad?
There are three big problems with this.
First of all, the industry specification just doesn't exist. And when you donít have proper specification and standardization, youíve got trouble. Youíve got bizarre conflicts, unstable systems, and worse. And because of that, youíve got angry end-users who will undoubtedly want someone to blame.
The DDR266 specification has only just reached an acceptable state. DDR333 still isnít anywhere close to being ready for mass-market. DDR400? Or is it PC3200? We donít even know what to call it yet!
Second, even if we had proper specification, thereís no performance or costing reason to push DDR400 right now, especially on AMD platforms (the Pentium 4 might actually be able to make some use of the bandwidth). DDR333 is more expensive than RDRAM right now, with virtually no performance increase over DDR266. DDR400 can only be more expensive, and offer even less performance increase. No oneís going to buy that. Just ask the Intel 820 and Rambus.
Third, the KT333 is only a few months old! VIA is entering down a dangerous path of releasing questionable, and for all intents and purposes, incomplete
chipsets, just to have something there before the Ďrealí chipset is ready. How many of you actually took the KT266 seriously, knowing the KT266A was coming? When youíre a business trying to promote yourself as a serious and reliable competitor, and both you and your customers know not to take half your products seriously because theyíre half-baked, youíre basically shooting yourself in the foot.
Why Is VIA Doing This?
As ridiculous as it sounds, I canít see any reason other than ďbecause we want to be in the driverís seat.Ē. Sounds preposterous, doesnít it? But why else?
VIA thinks it can introduce new memory standards on its own. Arrogant. Intel thought it could too with Rambus, and look at the grief it got. And VIA doesn't have anywhere near the level of presence Intel does.
A few years ago, that same Intel 820 put VIA in the driverís seat. What VIA doesnít seem to realize is that it didnít get put in the driverís seat because its products were the best, but rather because everyone else's were horrible. Big difference.
Real leaders donít try
What Should VIA Do Instead?
If VIA is serious about DDR400 (or DDR333 for that matter), it needs to sit down with JEDEC, memory manufacturers, and other chipset makers -- including Intel -- and make sure this beast is completely and entirely properly specified and feasible.
If the other industry leaders think itís feasible, then that means an extensive validation process is required. Thatís going to take time, and cost money. Welcome to the world of serious business. Do you think someone just decided PC100 would be fun one day, and turned it into the dominating standard for years to come with a snap of the fingers? Get real. It took years of collaboration and millions of dollars from tens of companies. VIA seems to want all the glory of pioneering new standards without putting in the work for it.
If DDR400 can be validated and properly specified to satisfy the performance and reliability standards of all parties, then
start talking about releasing chipsets. To do so beforehand undermines the entire process.
Otherwise, if VIA isnít really serious about DDR400, and this is just the latest cheap trick itís trying to pass off as quality to its customers, then Iíd say any hope the company had of being taken seriously is about as dead as the KT333.
For those interested, we've posted a follow-up to this article which can be read here