i am not understanding why a itanium processor would be better for a server. because, for example, the new itanium processor is 32nm, and consumes more power than a xeon. so, would the architecture help make the processor faster than a xeon nevertheless?
>>62169882
It doesn't, and it's being discontinued.
Itanium is a mess
>>62169882
>>62170067
>Itanium
Itanic you fuckers
>>62170019
source?
the itanium 9760 is being launched this year.
>>62170067
why? it is based on risc.
>>62170101
>it is based on risc.
No, it's not.
it's EPIC:
Explicitly Parallel Instruction Computing, which is based on VLIW, which is sort of an evolution of CISC
>>62170185
is it better than risc, then?
>>62170279
>is it better than risc, then?
instruction sets can be better or worse depending on they type of work they do.
RISC is good for math, floating point operations, that sort of thing
CISC can be better for procedural things like databases
EPIC could be better for procedural things, but it depended on compilers and software being written and architected to optimize for the EPIC /IA64 instruction set.
I'm not a computer scientist
>>62170081
And Intel have confirmed that the 9700 series is it.
>>62169882
It's for legacy reasons dipshit
>>62170279
The point is it attempted to simplify the processor by having more of the instruction level parallelism computed by the compilers instead of done at run-time by the chip. Unfortunately nobody managed to make a compiler that was terribly good at doing this. There's a knuth quote that such a compile time optimisation was intractable.
Itanium could perform well if you manually optimise and spend a lot of time rigorously testing potential optimisations, but then not a lot of people do this any more.
>>62170067
Itanium is a big fat mistake