> I'm still puzzled why Itanium didn't go any further.
It turned out nobody knew how to build effective compilers for them, so nobody bought them.
> And Intel just wrongly killed the project.
What's the point of a processor you can't program efficiently? What do you think Intel should have done? Just kept making them regardless of the fact nobody bought them?
> It turned out nobody knew how to build effective compilers for them, so nobody bought them.
It's a funny sign of those times. People really thought at one time that we were leaving a huge amount of potential instruction-level parallelism on the table. There was a DEC-WRL paper in the early 90s that said in the future (i.e., now) we'd be disappointed with 15 instructions retired per clock. But in the future we presently occupy we're all glad to have IPC of 0.8.
It turned out nobody knew how to build effective compilers for them, so nobody bought them.
> And Intel just wrongly killed the project.
What's the point of a processor you can't program efficiently? What do you think Intel should have done? Just kept making them regardless of the fact nobody bought them?