I am pretty surprised considering the demos Intel had only a week or two ago..... we'd concluded everything was going fine, and if HT really isn't in Penryn now, then surely that means the only new thing is SSE4?
I wonder if this is more a case of Intel dosn't want to release early as opposed to can't..... they have AMD where they want them right now, if they delay the launch of their 45nm stuff then they prolong the life of the 65nm chips - I imagine the last thing they want is to have a similar situation to clearing the Netburst stock once C2 came along.
This gives AMD a bit of breathing space if it turns out to be true. Does this also mean that Nehalem might be delayed? There wouldn't be much of a gap between Penryn and Nehalem if the original roadmap for Nehalem is stuck to.
It would be silly for Intel to give AMD more breathing room with Barcelona & Co. If they're pushing it back (truly, if this isn't just a rumor) it's for a very good reason.
That depends I guess - if they know something about Barcelona that we don't, then if they know C2 will hold its own against it, either as it is or by ramping up clock speeds on 65nm, then they could make more profit by holding it back, since they will have to lower their prices on the 65nm line as soon as 45nm comes out.
I can't understand what problems they could have run into considering how everyone was talking about the massive success of the recent demo :?
That depends I guess - if they know something about Barcelona that we don't, then if they know C2 will hold its own against it, either as it is or by ramping up clock speeds on 65nm, then they could make more profit by holding it back, since they will have to lower their prices on the 65nm line as soon as 45nm comes out.
I can't understand what problems they could have run into considering how everyone was talking about the massive success of the recent demo :?
Outside of leaks in AMD's labs Intel would have no way to know how K8L will perform.
Yeah I almost choked on my coffee this morning when I read that story. Same old Inquirer, spreading utter nonsense as usual. I suppose I'll file this one under the reverse hyperthreading story they ran.
I can assure you Penryn is not delayed. It is still scheduled for Q4.
That news is at a few other places too. It's the best thing Intel could do. They need to get rid of NetBurst before going to Penryn. It's going to be enough that they will be fire sale pricing most Clovertown SKUs to get more 2P volume.
Could it be that the three fabs won't be ready in time for Q4 '07?
Earlier stories indicated that D1D in Oregon and Fab 32 in Arizona were expected to ramp up production in "second half '07" while Fab 28 in Israel was always slated for "1H 08"
Maybe the two U.S. fabs won't be up and running in time to get Penryn to market by year's end? I don't know, just speculating.
Having a few 45 nm samples up and running is great, but that's a far cry from having mass production in place.
The tech seems solid, but there could be other problems. Labor disputes, work slows, shortages of building materials -- anybody know if all is fine with the construction and retrofitting plans?
Edit: I will now lend credibility to this rumor since digitimes supports it. Seems there may be a possibility of AMD matching (or exceeding) intel's performance for the last half of the year.
A little observation in the Intel release noted that the metal gates and high-k material would give an approximate 20 percent increase in drive current or more than.
Is that all?
I could have sworn that many, many other chip manufacturers at events such as IEDM were talking about double that figure? Otherwise why bother? Especially when straining has yet to reach its limits?
This could mean that Intel has rushed in dual metal gates and made some significant performance compromises. Only last year, Intel was touting more strain over high-ks and metal gates and even tri-gate structures for the 45nm node. I doubt very much that Intel's 45nm process will be anchored; rather it is going to float around a while before a better integration scheme can be utilized, perhaps at the 32nm node and the real benefits of this approach could possibly be seen
And rest of the article is here and is a very interesting read also.
So the delaying rumor might have some substance into it.
