Oh Dan, there you go again...
By using the notched gate process to etch gate sizes down to what would normally be the next major node (so they get .13 gates from their .18 process) Intel managed to keep from sliding too far behind Athlon in performance, but the resulting processors are fairly fragile, and they've also been "eating their future" by using up much of the benefit that otherwise would be gained when moving to a copper .13 from an aluminum .18 process.
Actually the "notched process" gave channel lengths of something around 100nm which was the industry standard for 0.18 micron. The industry standard for 0.13 micron is 65-70nm. At the newly renames 90nm node (formerly 0.10 or 0.09 micron), the standard will be 45nm, at 65nm it will be 32nm, at 45nm it will be 22nm, and at 32nm (in 2009) it will be 16nm. Intel has demonstrated transistors down to 20nm.
PS - PIII went from 1.1GHZ on aluminum .18 to 1.26GHZ on copper .13 for a gain of only 15%!
Ummmm... 0.13 micron is just ramping, and 1.1GHz on 0.18 micron is relatively recent. It certainly wasn't released in 1999. Tom overclocked the first Tualatins to 1.5GHz, and Intel has demonstrated 3.0 and 3.5GHz P4's on 0.13 micron (it topped at 2.0GHz on 0.18 micron in case you lost count). Looks like the 0.13 micron process is perfectly healthy to me.
You seem incredibly ignorant of basic process technology as well as technology cycles, and quite desperate to prove some point which is not backed up by fact. If AMD hadn't blown their wad on channel lengths, they wouldn't have to be spending their way out of a performance deficit (SOI). Unless AMD can start inching its ASP's up dramatically, the combination of Intel at 300mm and AMD on SOI will bankrupt AMD. |
