The biggest thumb-in-the-eye problem for Itanium was that it required new compiler technology (for EPIC technology) which was never developed. As I understand it, most speed improvements from Itanium came from other technologies like: moving to DDR3 memory, die shrinks, multiple cores, all which would have all happened had Intel stuck with HP's super-scalar RISC (PA-RISC, DEC-Alpha, etc). Okay so now the "big race" is down to three horses: Itanium from Intel, POWER from IBM, SPARC from Oracle/Sun. Meanwhile, x86-64 gets more powerful every 18 months and many people wonder if the big three horses are on the verge of becoming irrelevant. With Oracle pushing SPARC, Intel and HP only have one option: Give Oracle's hardware business a spanking by dropping the price of Itanium hardware and software.
p.s. server sales over the past 12 months show HP-Itegrity (HP's brand for Itanium platforms) with at least 80% market share in the Enterprise Server Category. Having Oracle withdraw enterprise's favorite database product will cause this number to drop. Other than a full-scale price war with Oracle, HP doesn't have any other choice than to send Oracle payola to save Oracle on Itanium (cutting Oracle in on Integrity profits)
the best way to understand iAPX 432 is simply "the anti-RISC" - it's almost as if they looked at everything Cocke et al were learning, and chose the opposite.
anyway, ia64 was very usable, and sometimes even competitive in performance. the problem was that x86_64 was so much easier to get along with, vastly cheaper, and soon ubiquitous. the story for ia64 was basically over as soon as Intel gave in to AMD's x86_64. proponents still argue that ia64 has unmatched RAS, but the fact is that modern service architecture is based on having lots of machines, rather than one gold-plated megabox with hotswap raid1 dimms. there's still a market, tiny in volume, for mainframes, though. I don't think that's enough for ia64 to survive on though.
You wrote: "Several years ago, Intel rolled out Itanium for high-end servers. The product was late and has been a major bust. Perhaps it is Intel's biggest mistake ever."
Even worse than the iAPX-432? I don't think so.
Blog Doing Math in FPGAs Tom Burke 15 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...