Intel Creates 45 Nanometer Architecture

Rob Updegrove 25 Jan, 2006 at 2:39pm ET Article published in Tech

Comments

1. 
   25 Jan 2006 ~ 3:16pm jradmin Lol and it will probably be slower then AMD's 90nm cores.
2. 
   25 Jan 2006 ~ 11:27pm Leonardo

   it will probably be slower then AMD's 90nm cores

   Yes, probably. But at least it should be an end to Intel's last two disastrous years of self-inflicted wounds - Itanium, Smithfield, Presscot...
   
   The corporate computer buyers and server makers have finally awoken to the high power consumption levels of 2004/2005 Intel processors. 65Nm should at least prove to be a legitimate stopgap to some of Intel's problems. AMD still has a lot of momentum going. Intel will lose even more market share to AMD before they stabilize and produce first rate technology.
3. 
   26 Jan 2006 ~ 10:22am Newer processes will only have more problems with current leakage and that was the main problem that Intel, AND AMD for that matter, had with the 90nm process. The smaller and smaller you go the more problems with current leakage. The voltage was lower on the 90nm CPUs and the overall wattage was higher which means the current jumped way up. Multicores should help alleviate this problem with reduced frequency and going wider not longer.
4. 
   26 Jan 2006 ~ 11:41am Qeldroma

   (Guest) wrote:

   Newer processes will only have more problems with current leakage and that was the main problem that Intel, AND AMD for that matter, had with the 90nm process ...

   
   And that is only a single issue. There are heat, voltage threshold, materials, noise, cross-talk, and wavelength issues involved as well. (I personally think the last is where the buck finally stops with the current designs). However, why publically publish it and go to production with it unless you can manage it? They've fought these battles before and have won so far. However, Guest, you are absolutely right in that no matter what the process, we are NOT going to see any more dramatic jumps in frequency (which, BTW, there is a little more room for), but in more features- which I see could even be going to "on-chip" RAM, GPU, etc and not just cores.
   
   But right now, from what I've seen of Conroe and other cores on the board, Intel's more immediate and critical problems revolve more around core design and not frequency or process. It's more like "They can't seem to get their inter-core communications out of their FSB" type issues. AMD has been winning the game of getting more bang/cycle, and that seems to be the trend Intel has to beat. Imagine if today AMD could match Intel cycle to cycle! What's to say they won't in a year or so? It's about the only edge Intel has right now.
5. 
   26 Jan 2006 ~ 12:15pm Leonardo

   Imagine if today AMD could match Intel cycle to cycle! What's to say they won't in a year or so? It's about the only edge Intel has right now.

   And that is an edge, but only an edge because ignorance is still widespread, believing CPU frequency is in and of itself an indicator of performance. AMD's PR rating system, despite all the criticism it has received, created at least some awareness of the concept of work per CPU cycle. That was start to educating the users (private and corporate). When Intel dropped their naming convention that was based on CPU frequencies, this served to bolster AMD's performance rating arguments. When the business world found out how expensive (power consumption) it is to run much of the current crop Intel processors, AMD again received a boost. Alas though, the megahertz myth is still going strong, but is waning.