Itanium 3 410 mil/transistors

[-=DVS=-]

Whoa this one is giant CPU :eek:

The 1.5GHz "third generation" Itanium processor, presented by Intel, will speed along at 1.5GHz and have an on die cache of 6MB, and be 374mm 2. Intel will say it has 410 million transistors and will be implemented on a .13µ (micron) package using six level Cu interconnects with FSG dielectric, dissipating 130 watts.

Link to Source (http://www.theinquirer.net/?article=7413)

Its somewere 3 times more then NV30 or R300 GPU,
Several times more then any other current CPU :eek:
With 6 megs of cashe what a monster :p

[druga runda]

And will it cost 5000 USD for one CPU, yes this is Opteron competition :afro:

Or when is Intel going to sell these for the desktop market :p

Itanium 3 Celeron @ 1ghz with 2 MB of cache for 60 USD - runs twice as fast as PIV 4ghz hyperthreaded :D Hey GHZ don't matter folks, who cares that those opterons run at 3 ghz we have better IPC :afro:

[SavagePaladin]

As the businesses are already saying "It's not worth the investment" I don't think thats going to change. For a loooong time.

[druga runda]

What a strike

the 6MB Itanium III will cost $4,226 at release :D

A bit short of my $ 5000 prediction but close enough.

here is the Inq link
http://www.theinquirer.net/?article=7454

I hope it is finally the time Intel starts digging its own grave. They were on the top of the game for too long. Make somme room for loss making AMD and let them have some profits

:angel:

[Steppy]

Somebody better go look at some facts...even though it has wrongly been compared to it, Itanium and Opteron are NOT competition for each other. 5000 for a processor where Itanium is aimed at is CHEAP. I can't wait for final benchmarks to come out and FINALLY show people the opteron is aimed for Xeon's NOT Itanium(which CRUSHES it in performance). You won't ever see an IA-64 processor marketed at home because the ISA is NOT suited for a desktop machine(it's highly parallel and doesn't do the random tasks desktop users do very well).

[Steppy]

Originally posted by SavagePaladin
As the businesses are already saying "It's not worth the investment" I don't think thats going to change. For a loooong time. This is because the big iron market is slow to change and Sun and the like are LONG entrenched in this market...it's not worth it to switch over to a new architecture, and this market takes a long time to break into.

[SavagePaladin]

I agree they're not competition, but I don't think Itanium, with its huge power draw, huge cost, and need for new software is much of an option for anyone.

[Steppy]

It's not a huge cost...itanium is by far one of the CHEAPEST solutions in the big iron market(even though its better performing than most of the other solutions in the big iron market, the cost of moving over to a new architecture is prohibitive...but the hardware itself is VERY cheap(a 4-way I2 system I believe is cheaper than a one way alpha). It's because of the need of new software mostly

[druga runda]

Well I don't know much about high end server costs/chips (except that they can run into millions :D ); but there are those year in year out rumors or ideas that Itanum will (or at least should) become the 64bit desktop processor one day. I could not really believe that Intel would spend upwards of 10bn on R&D for something that will not run on the desktop, or just compete in the business sector. It sounds like a big flop, still its days are far from numbered. After all the processor is just a part of an high end machine, and the integrator makes the most of it, like HP who these days after acquiring Compaq builds alphas and make servers, or IBM or Sun etc... in Intel's case they are just a processor company, to spend 10+ bn on seems a bit excessive.

Anyway there was at least an idea around about a desktop Itanium, however this 6MB cache 410 million transistor beast for $5000 just doesn't seem to cut it. Nor can I see any other Itanium offspring that could somehow impact the desktop PC.

While Opteron will not only strike at Xeon dual processor sector but they can be scaled up to 8 way systems (and upwards I believe) for significantly less cash, with x86 compatibility. Perhaps a bit low on the performance, but if performance is what you are after IA64 still doesn't beat Alphas, even though it seems it will become Alpha's ancestor in a couple of years. But that is about it from what I could read. Still they will have at least IBM's Power5 to comete with in that sector when the time comes, with Opterons being good enough for more "middle range" operations than any "low end" significantly cheaper processor before so it is not really a free rule. (unless AMD goes bust before of course :D )

[StealthHawk]

Itanium won't be in desktops systems for 2 reasons
1) it runs legacy 32bit code poorly
2) it's too expensive

[Steppy]

Originally posted by druga runda
Well I don't know much about high end server costs/chips (except that they can run into millions :D ); but there are those year in year out rumors or ideas that Itanum will (or at least should) become the 64bit desktop processor one day. I could not really believe that Intel would spend upwards of 10bn on R&D for something that will not run on the desktop, or just compete in the business sector. It sounds like a big flop, still its days are far from numbered. After all the processor is just a part of an high end machine, and the integrator makes the most of it, like HP who these days after acquiring Compaq builds alphas and make servers, or IBM or Sun etc... in Intel's case they are just a processor company, to spend 10+ bn on seems a bit excessive.

Anyway there was at least an idea around about a desktop Itanium, however this 6MB cache 410 million transistor beast for $5000 just doesn't seem to cut it. Nor can I see any other Itanium offspring that could somehow impact the desktop PC.

While Opteron will not only strike at Xeon dual processor sector but they can be scaled up to 8 way systems (and upwards I believe) for significantly less cash, with x86 compatibility. Perhaps a bit low on the performance, but if performance is what you are after IA64 still doesn't beat Alphas, even though it seems it will become Alpha's ancestor in a couple of years. But that is about it from what I could read. Still they will have at least IBM's Power5 to comete with in that sector when the time comes, with Opterons being good enough for more "middle range" operations than any "low end" significantly cheaper processor before so it is not really a free rule. (unless AMD goes bust before of course :D ) 32-bit capability is of zero importance in the Itanium market. The ONLY time I heard of Itanium being used in desktops was in an article written back in 95 or 96, back before a whole heck of a lot was known about the epic architecture. The architecture is HORRIBLY suited to being a desktop machine, and Intel and HP knew that when they designed it. It's for going for the high profit big iron market. Actually, Itanium DOES beat Alphas in specFP, it beats pretty much everything else out there. But this market doesn't change very fast, which is why you seen slow penetration into the market for ANY new player even one as powerful as Intel. The cost of the system is a drop in the bucket, changing over all your software and retaining usability for all your data is what is expensive. Itaniums bread and butter was scientific apps, and likely always will be(this thing is a pure number crunching beast). We're VERY early in the EPIC architecture age...basically like about 1982 or so if you wanna compare it to desktops. If and when Intel decides to go for a 64-bit desktop CPU, you'll see a much more clock scalable architecture with a much less parallelized ISA(could be x86-64 if AMD gets decent support under its belt, will not be anything x86 if AMD doesn't). Like I said, 99% of the time you'll never see Opteron competing with Itanium(they will be some cases where an opteron server IS a better solution of course, but it will be an exception far from the rule). This thing likely could and hopefully WILL hit the Xeon market like a excuse the pun "sledgehammer". But in the big iron market its a rubber mallet.