If you look at what was said at the AMD Analyst Day conference, well, the issue isn’t whether AMD is going to be running behind Intel the next couple years. The issue is whether they’re going to be running at all.
From our perspective.
What is happening is that what is and will be a good answer for you and what is a good answer for your Sixpack friend are likely to diverge, and stay that way.
I’ll put it this way: It will probably make little to no sense for overclockers to buy an AMD processor for the next few years. The 4X4 trails Intel’s quadcore (and 4X4 really looks like a server system in disguise anyway). AMD won’t provide retail 65nm chips for a few months, and if you look at the preliminary overclocking results here and here, this is no great loss.
The next generation of 65nm chips probably won’t be any great shakes either, again, for us. AMD apparently thinks it can get the IPC on these chips roughly equal with today’s Conroes.
That’s nice, but by the time they do that, Intel will be debuting its 45nm chips, and even if they don’t improve the IPC, their default speeds will be considerably higher than AMD’s, and O/C speeds probably even more so.
And AMD won’t get to 45nm for at least a year after that, and Fusion chips won’t show up until around 2009.
Frankly, from our perspective, things look pretty bad for AMD. However, ours is not the only, or most important, perspective.
In a sentence, AMD no longer terribly wants to sell to us. They want to sell to your Sixpack friends, and if you look at it from that perspective, things look a whole lot better.
It doesn’t mean squat to a Sixpack with a Dell that a Conroe processor can overclock better than an X2, he isn’t going to be doing that. If he’s even aware of the IPC difference between the two, he’ll just buy a 2.2GHz X2 rather than a 1.86GHz E6300. Odds are he won’t, so he’ll buy a 2GHz X2 if it’s cheaper and be none the wiser for it.
Look down the road, and the same pretty much applies. What matter greatly to us won’t matter much or at all to a Sixpack, AMD will be good enough for him.
Look at servers, and the prospects of dual-quad cores married to a faster Hypertransport ought to keep AMD competitive in that market for a while, too.
When 2009 comes, then we’ll see Fusion and the difference between Intel and AMD will become even more a matter of comparing apples and oranges.
And that’s probably just what AMD wants.