I'm anything but thrilled with Intel's concoctions like RDRAM, MTH, and the P4/Willamette. However, their Itanium draws a lot of my sympathy. Why? Because it's the first serious attempt to leave the obsolete, doomed, sinking x86 architecture (at least for common PC's).
The irony, of course, is that this chip was originally intended to debut in 1997 according to this recent sneak preview. As a result, its working samples are themselves severely outdated by newer x86 chips. Which is obsolete and sinking now? Enter the term "Itanic."
But I'd much rather see Itanic being redesigned to succeed. Does anyone know if Intel is still trying to make its "next gen" IA-64 a viable future platform? Just think of what you can do with a 64-bit CPU that has 128 general-purpose registers (x86 CPU's sport 4 32-bit ones! ). Theoretically such a CPU should be miles ahead of x86 by design alone (even with a similar fab process). Obviously apps need to be designed/compiled in radically different ways to use the parallel architecture. The speed would also allow software emulation of x86 programs. This means you'd be able to run Glide and Windoze games in emulation under Linux. The architecture could also be streamlined without the legacy crap that makes PC's so painful to configure (and slows them down). No more IRQ and instability nightmares. Hey, it's OK to dream sometimes.
Can anyone supply a reality check? What's really happening in the foreseeble future?
The irony, of course, is that this chip was originally intended to debut in 1997 according to this recent sneak preview. As a result, its working samples are themselves severely outdated by newer x86 chips. Which is obsolete and sinking now? Enter the term "Itanic."
But I'd much rather see Itanic being redesigned to succeed. Does anyone know if Intel is still trying to make its "next gen" IA-64 a viable future platform? Just think of what you can do with a 64-bit CPU that has 128 general-purpose registers (x86 CPU's sport 4 32-bit ones! ). Theoretically such a CPU should be miles ahead of x86 by design alone (even with a similar fab process). Obviously apps need to be designed/compiled in radically different ways to use the parallel architecture. The speed would also allow software emulation of x86 programs. This means you'd be able to run Glide and Windoze games in emulation under Linux. The architecture could also be streamlined without the legacy crap that makes PC's so painful to configure (and slows them down). No more IRQ and instability nightmares. Hey, it's OK to dream sometimes.
Can anyone supply a reality check? What's really happening in the foreseeble future?