Any studies show IA64 migration to be benefiicial to overall computing performance after the migration?

MadRat

Lifer
Oct 14, 1999
11,971
288
126
IA64 benefits over IA32 are so wide and verying that I cannot see IA32 in any form being around in five years. AMD64 is pretty popular as far as being the near term solution to future proofing computing performance in the enterprise. I was wondering if there were any IA64 studies that actually shows a capacity or performance comparison between pre-upgrade and post-upgrade systems. If a system is roughly the same cost to maintain yet 40-50% overall higher performance then its hard to justify the move to AMD64. If the IA64 system has higher maintenance cost yet still performs relatively better then it still would be hard to justify a move to AMD64, since both systems require OS changeouts. We all know that eventually we run into a decline the in the cost/performance differences where AMD64 or IA64 would never make sense, and IA32 maintenance is best. I'd like to here a scientific analysis of whether or not IA64 is worthwhile for future proofing. It doesn't need comparisons necessarily to AMD64, but it would be nice to see that, too.

My personal opinion leans towards IA64 being delayed for mainstream users as clockspeeds for desktop CPU's continue to escalate at their current rate. However, in the next several years we'l reach a point where ILP (instruction level parrallelism) is better than ramping raw CPU clockspeeds and then makes IA64 adoption inevitable. Intel seems to be hinting this way with their large swath of Itanium 2 processors due to hit the market this and next year. Intel may be able to get consumers attached to IA64 with a seamless switch from x86-32 by using hardware-level IA64 emulation in thier Prescott and beyond processors. Just a hunch, but Intel's roadmaps tend to lean towards a unified outcome of IA64, which makes people have to decide whether short-term AMD64 migration is even a worthwile chore.
 

DaveSimmons

Elite Member
Aug 12, 2001
40,730
670
126
Desktop or server?

Considering that office workers don't generally need more than about 1 GHz I think you'll have a hard time finding any study showing benefits from performance gains in the over-3-GHz range. Niche desktop users (graphic designers, CD, engineering) maybe.

Also, since Windows 64 is still beta, any studies would probably need to be linux-based.
 

MadRat

Lifer
Oct 14, 1999
11,971
288
126
I'm looking at the mid-range server, mainly the range that use 2-8 CPU's, not the upper end of the mid-range and well short of big iron. Its not for my own personal benefit, its more for just general questions on the viability when talking the strategic positioning of an enterprise environment over a 5 year plan.