With the next Itanium chip, Intel has abandoned a feature it once banked on but that never proved successful. Circuitry to let Itanium run software for x86 chips, such as Pentium and Xeon chips, is not present in the forthcoming ‘Montecito’ processor, according to the 176-page reference manual (.pdf) for the chip published this week. Update: El Reg has more on this story.
This is good. get rid of the x86 cruft in the cpu… allows more room for other things..
Its about time. I don’t know of anyone that ever considered running x86 on the itanium and now that x86 64-bit has arrived it is no longer needed.
why invest more money producing a CPU that nobody will buy, because everybody will be buying AMD’s opterons instead.
I thought this was known since Montecito was first described in detail, which I guess was in 2004. I believe Intel has had an emulator available since about that time that is actually faster than Madison’s x86 unit, and I think with Montecito it was decided to make this the only route.
Besides, this makes sense given that it really only needs to be available for Windows and maybe Linux to be useful — no x86 HP-UX binaries out there
Anyway, very good move.
It was removed because it’s actually *faster* to do the emulation in software.
True, hence the reason I can’t get over why there was such a big thing made – would have thought it been logical as the software emulation had been mentioned some time before.
As for over all design – it should mean in a reduction in price – but due to Intels insistance of not allowing normal distrinbution channels from selling Itanium motherboards and processors, they can’t get the volume up as to produce the economies of scale and reduce the cost to the end customer.
Intel offered me a ‘developer workstation’ on a ‘lease’, my response, ‘no thank you, I’m not going to start throwing my money into a bottomless pit – either give me the option to purchase a kit which allows me to configure and set it up as I like, or forget about it’ – to which the Intel rep tried to ‘hold onto me’ by saying, ‘something will be coming soon like that’, over a year and a half later, nothing.
Why doesn’t Intel just admit they don’t want Itanium to succeed, because it seems they do everything humanly possible to make it a bloody failure.
They don’t sell very well, server based on it cost alot more and for what?
You are much better off using Xeon (EMT64) or Opteron CPU in your servers. You can almost have 4 Opteron CPU for the same price as 2 Itanium….
If you realy need something more powerfull, get a SUN with Sparc CPU or IBM with Power5.
the negative return on investment for itanium is breathtaking even for intel. no, markets for this chip are not going to materialize. no, people will not wake up to its value “real soon now”. no, no one really cares if it is slightly faster than brandX.
its DEAD. sales of this chip are measured in the same way prisoners mark the days – scratch marks on the wall.
even intel can’t kill x86.
Of course they can. If they start supporting x86-64 in their CPUs more people will switch, so they can kill legacy x86.
Because its the fastest processor available. Look at the top500 list. And ia64 is very progressive and powerful architecture, problem is the lack of good compiler for it. But AFAIK Intel is working hard to improve the compiler, and they hunt and buy all professional compiler developer teams they can find around the world. In 2005 IA64 compiler got 30% performance rise.
Browser: Mozilla/4.1 (compatible; MSIE 5.0; Symbian OS; Series 60;452) Opera 6.20 [en]
>> Because its the fastest processor available. Look at the top500 list
to be blunt, no one cares. nor should they. in fact if someone is using the top500 list to form opinions on business hardware, they should be relieved of this duty.
no one is going to buy a cpu because it is used in esotertic high-end, one-of-a-kind clusters, because they know this is meaningless to their implementation.
with very few exceptions, business computing and high performance computing went their separate ways long ago. compatibility with current operations, vendor support, low maintainence, low power consumption, low heat dissipation, commoditization (can i buy these types of systems from someone else should dell or hp or ibm piss me off), are all far more important these days (among other reasons) than flat-out cpu performance, which is not to be confused with *system* performance (throughput, i/o, etc), which is also rarely dependent on cpu per se anymore.