Originally posted by olePigeon:
Intel is little-endien only while Moto/IBM are both big and little.
Either way it shouldn't be such a huge deal to port the Carbon API and its apps to IA-64.
The other interesting thing to note of course is that IA64 like the Transmeta Crusoe has the potential to support more than one processor architecture. In fact in this arstechnica article, it is suggested that Transmeta might even do that. i.e produce a PowerPC variant of its processor:
http://www.arstechnica.com/cpu/1q00/.../crusoe-5.html
I'm surprised Intel hasn't purchased Transmeta if only for the 'code morphing' technology which would make an excellent solution to the Itanium requirement of supporting two legacy instruction sets: IA32 and PA-RISC. (This would be instead of budding on *two* additional processor cores onto the chip: Itanium is really *3* processors in 1).
So Intel could if so desired produce a PowerPC compatible version of IA64.
The promise of VLIW is at least as radical as RISC was when CISC was the (only and) dominant microprocessor design philosophy around.
Transmeta is certainly proving that.
In fact, coupled with Intel's superior fabrication technology, they could *really* up the Ghz and performance of a simple IA64 VLIW/EPIC core (by replacing the Pentium and PA-RISC cores by a software 'code morphing' layer). If there is still chip real estate to spare, stick in a second VLIW/EPIC core on the same chip. Or perhaps a SIMD unit?
Top