For overclockers, up to now, if you wanted to look at a Hammer board, you had to look at nVidia, because Via had never, ever had a PCI/AGP lock on anything.
That’s the kind of default choice you’d rather not have.
So, quite to my pleasant surprise, Via has finally decided they want some of our money, and included a PCI/AGP lock in their upcoming K8T800 Pro.
Unfortunately, most of the reviews I’ve seen of this board dissed overclockers. They all talked about the PCI/AGP lock, but they didn’t test it.
Most said they “didn’t have the time.” Yeah, right. How long does it take just to see if it works?
Just realize that such folks put us at the very bottom of their totem poles, and if they “don’t have the time” to test, then you don’t have the time to read them.
Fortunately, there was at least one review that decided to tell you what you really wanted to know.
The results themselves aren’t all too important since this is just a reference board. We’ll just have to see what magic the Taiwanese wizards will come up with for boards you’ll actually end up buying.
What is important for now is that there’s going to be two major contenders for your future Hammer mobo money rather than just one.
That’s good, very good.
Now Or Later?
Both Via and nVidia will put out AGP-based socket 939 boards initially. A few months after that, we’ll see PCI-Express based boards (hmmm, I guess we’ll need a PCI-Express lock of some sort, too).
Which to buy? Well, it really depends on what you’ve done in the video field lately.
If your card is relatively new and/or you don’t plan on replacing whatever you have for at least another year, AGP would appear to be the way to go.
If your card is showing signs of age and/or you plan on replacing whatever you have within a year, PCI Express would be better for you.
We say “about a year” because we suspect that 12-15 months from now, AMD will go over to DDR2, and doing that will require a new socket, but PCI Express won’t be going anywhere.
It seems like the proper approach for most to switching over to desktop dual-channel Hammer is either “adapt very early or very late.” This is why we think the “price it high and watch sales die” approach AMD is likely to take is pretty stupid.
Get a reasonably priced socket 939 out there (by reasonable, I mean $150-200), and lots of people (by that I mean a few hundred thousand, just enough to get most of that unused capacity at Dresden working) will buy one early, top it off with a 90nm chip when they get relatively cheap next year, and be ready for more.
Price socket 939 chips at $400 or more come June, though, and those people aren’t going to pay $400, they just aren’t going to buy anything, and if you don’t get into their price range with a 90nm until a year from now, they’ll probably wait for the DDR2 version. If they haven’t settled for your cheapie Paris by then, that is.
To me, that’s found money, and I’m very suspicious about any company that would turn up their nose at that. Especially when the competition is hurting.