Going to 64bit will allow the use of larger integer values but not more "accurate" ones. Current 32bit machines all support 64bit floating point (numbers with decimal points) and the move to 64bit won't increase that.
yeah, since the 486DX with the onboard math coprocessor, we have had double length floating point numbers in the x86 world. i remember going to the store and thinking about buying a math coprocessor for my 386. man, those were the bad old days.
Key is more bandwidth, more instructions per cycle. Faster. But the transition may be a problem, unless they can make a 64-bit processor that can run 32-bit program well, we'll be seeing problems here and there. Should be itneresting to watch
<< The Intel Itanium is 64bit and it can run 32bit code. Of course, the way they did it was by putting a p3 core on the itanium die. >>
I wouldn't think it would be as easy as slapping a P3 core on it. Completely different architecture on my mind, though Intel and AMD have been producing chips using the same architecture for decades. This may be Intel trying a hand on making another variation.
bandwidth is NOT related to "bitness" of a cpu. you could, if for some reason you wanted to, design a 256-bit CPU (256-bit registers) that only uses an 8 bit bus and serializes everything (uses 32 cycles to finish addressing / sending or receiving data).
<< Text I was under the impression that the Itanium translated the instuction >>
I'm pretty sure that I read that the Itanium doesn't translate instructions but instead has some sort of p3 core on it. It wouldn't be so hard to fit a P3 on an itanium since the itanium die has a ridiculous 320 million transistors on it.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.