2005-10-24 Intel confirms Itanium schedule screwed up

Link to story (The Inquirer)

Other news this month

Related Roadmap Items

2006-07-18 Intel Montecito Dual core IA-64 processor with Switch on Event MultiThreading (SoEMT), (like Niagara) built on a 90 nm process. Expected to improve on Madison 9M's performance by 2.5 times. To include 2 * 1MB of L2 cache, 2 * 12MB of L3 cache and around 1.7 billion transistors.
Consumes "only" 100W. (Foxton was to clock it up to 2.0 GHz when it was running cool enough, but was canceled.)
The 2.0 GHz / 24 MB versions are delayed until Q2 2006.
Will be named 9000 (9015, 9020, 9030, 9040 and 9050). The 9010 is a single core version.

2006-H2 Intel Montvale Itanium processor, succeeding Montecito and preceeding Tukwila. Earlier said to be a 65 nm major upgrade, but now rumoured to be a 90 nm chip with just some extra cache.

2008 Intel Tanglewood / Tukwila IA-64 processor originally said to have eight (plus one spare) cores per die and 16-32MB of cache to follow after Montecito. Appears to use parts from a follow-on project to the now canceled Alpha EV8. Said to arrive in 2006 or 2007 and offer "at least seven times the processing" power of Madison. Other rumours claim that HP wants their advanced math libraries on the die in silicon and that it might include a vector engine.
The 8+1 core version (designed in Hudson) is now said to be replaced by a 2-core version designed in Fort Collins.
Recently said to be renamed from Tanglewood to Tukwila due to copyright reasons.
Should use the new CSI bus and socket like Whitefield). Latest info indicates it will include 4 cores with 6x4 MB of L2 cache and an on-bord FB-DIMM memory controller..

Compare Roadmap Items