wbmw,
It is not an unfair assumption, though. AMD is trying their best to be a PR machine, and they routinely show off products ahead of time to make it seem like they are ahead of Intel in every category under the sun (and they have some people fooled with this tactic).
Really? When it comes to demonstrating future stuff (to fool people, as you say), I would say that Intel has been doing more of that (with all of the overclocked P4s demonstrated at IDFs).
BTW, why would AMD try to fool people that they are ahead, when all you need to do do is check what's offered for sale. Outside of Pentium M, AMD is anywhere between marginally to way ahead across all of the remaining segments.
As far as the pipeline It's kind of a projection, and IMO, AMD is being overly secretive in this department.
Now, keeping in mind the grey areas between process transitions, nothing will preclude AMD from having a 90nm flow with aggressive gate lengths and calling it a "65nm" shipment, and no one would know the difference.
Until Merom ships, 65nm for AMD is only an issue if they run into production capacity on existing 90nm process (in Fab 30), and if there is no additional output from Fab 36 (90nm?) or Chartered. Intel may narrow the gap with 65nm Netburst (but will not gain lead), and AMD is still nowhere in notebook segment.
The end result is that I don't expect AMD to have any more striking power or performance improvements for another 18 months, whether or not they stick to their supposed schedules with a PR product release.
I don't see any performance breakthroughs either. One or 2 speed grades on single and dual core is probably the max we can expect. They may add a core or 2 within this timeframe, and there may be similar (small) gains on the power consumption front, but again, no breakthroughs.
AMD does have a new, low power core on the roadmap for the mobile segment, but last time it showed up on the roadmap, it was still a little down the road, so I would wait for the fall analyst meeting presentations for an update. It may end up being outside of the 18 month window that you mentioned.
I think it is from Boston design center. I hope these guys don't work on Sun schedule... Then, there is a new core for mainstream market scheduled for 2007, which, I am pretty sure, "borrows" some of the performance enhancing ideas that went into Banias line of processors.
If you want to hold me to this, let's just see when they are able to either A) scale past 2.8GHz in a dual core product (which I believe will require 65nm), or B) introduce a mobile dual core mobile product (i.e. 35W or less) that scales higher than 1.8GHz. My guess is that we'll see these products at the tail end of '06, or more probably, early 2007.
Sounds reasonable. I would be surprised to see more, and I would be surprised if 65nm, right of the bat, was significantly ahead of mature 90nm.
Joe
Market Data 
Markets 
AI
