Linked by Thom Holwerda on Wed 1st Mar 2006 22:52 UTC, submitted by Robert Escue
Intel "On January 26th, 2005, the Itanium Solutions Alliance (Intel, Hewlett-Packard, SGI, NEC, Hitachi, Bull SA, Fujitsu, and Unisys) proudly announced that its members had banded together to invest $10 billion over the next four years to improve Itanium's features and functions and strengthen its market position. This huge investment in this architecture at this time forces us to ask a lot of tough questions about the wisdom of this action... [.pdf]"
Thread beginning with comment 100687
To view parent comment, click here.
To read all comments associated with this story, please click here.
Member since:

Sun can make a new proccessor with 8 cores and 4 threads per core, that trounces x86 and amd on most non fp tasks...

Dude, it's not just the numerical number of cores per CPU. SUN T1 was designed for massive multithreading apps and each core is based on obsolete in-order execution mechanism. It's NOT a number-crunching CPU. Even more, if application are not carefully written ( access to single global variable will serialize execution of ALL hardware threads running on each core), most apps won't have any benefit.

Itanium's EPIC-style VLIW design eats every single RISC/CISC for breakfast. It is very unfortuate that the original Itanium's engineers decided to use high-latency cache that clobbered the whole execution pipeline. If Alpha was killed by corporate greed, IA-64 will be killed by market inertia.

OTOH, AMD64 has the best price/performance ratio than any other CPU. I guess x86 curse will float around for at least a decade. I'm so sick of x86 and CISC vs. RISC debate, and would really like to see IA-64 on desktop ;)

Reply Parent Score: 5

Chreo Member since:

Itanium's EPIC-style VLIW design eats every single RISC/CISC for breakfast.

...for the task it was designed. Nothing more, nothing less. It is not too hard to kill the performance of itanium since it relies so heavily on compiler optimisations and the huge caches. Itanium is, by design, In-order and in far too many ways feels a bit "yesterday" than fresh and innovative.

The field of excellence for itanium is FP-math. Integer calcs are not too impressive.

IA-64 will be killed by market inertia.

Partly, itanium will be killed because the enormous investments have not paid off and the desktops promises have utterly failed to materialise (wasn't 2004 supposed to be the year of a good desktop itanium cpu?). Itanium and the NetBurst core was born out of the idea that transistors get faster and that the cooling ideas of yesterday would work in the future. We know what happened to NetBurst... Where is the desktop itanium that does not require a personal powerplant to run?

Reply Parent Score: 3