Sign in with
Sign up | Sign in
Your question

Itanium III

Tags:
Last response: in CPUs
Share
June 19, 2002 7:46:05 PM

<A HREF="http://story.news.yahoo.com/news?tmpl=story&ncid=70&e=2..." target="_new">Click</A>

Don't know if it's going to be named Itanium III. But it's Alive and running.

KG

"Artificial intelligence is no match for natural stupidity." - Sarah Chambers

More about : itanium iii

June 19, 2002 7:52:48 PM

Quote:
Intel President Paul Otellini, speaking at a press briefing at the Securities Industry Association Conference here, said Madison is "alive and wiggles"


:eek: 

Quote:
Otellini said Intel is not planning chips that can handle both 32-bit and 64-bit code.


I have a hard time believing that.

Quote:
Nevertheless, Otellini said Intel has reserved patents to extend its 32-bit Pentium and Xeon chips to handle 64-bit code.


Interesting...

<font color=blue>Hi mom!</font color=blue>
June 19, 2002 8:03:03 PM

Reading this article will show you why Intel is the dominant chip maker on the market. They cover all their bases. With McKinley, Madison, Deerfield and Yamhill they are offering a comprehensive product line for the high end server market. The wrench in their plan will be if AMD's Opteron can perform at or above McKinley levels and is bilingual because we all know it will be cheaper then McKinley.

HULK SMASH!!!
Related resources
June 19, 2002 8:14:18 PM

Intel isn't stupid, Yamhill will be bilingual and available next year if they not what is good for them. AMD already has the edge just due to the fact that Opteron is bilingual which to me is incredible.

I believe Intel trying to force people towards 64bit programming is actually a good thing but I think they are going about it in the wrong way.

HULK SMASH!!!
June 19, 2002 8:17:40 PM

Will someone come up with a better naming convention that simply adding XP or a number after a CPU for the next generation of them? At least one of the hammers will be named Opteron.

Are we out of good CPU names?

It can be said that smoking is one of the leading cause of statistics.
June 19, 2002 8:17:57 PM

Will someone come up with a better naming convention that simply adding XP or a number after a CPU for the next generation of them? At least one of the hammers will be named Opteron.

Are we out of good CPU names?

It can be said that smoking is one of the leading cause of statistics.
June 19, 2002 8:31:41 PM

I can see it know.....
On a quiet day in 2050.
Intel announces the Pentium XXXV
Ill be telling my grandkids that back in my day we used to have processors called 486s....and theyll think im just senile....
June 19, 2002 8:36:54 PM

"To get the full performance benefit out of Itanium, customers need brand-new software written for the chip. Sources have said that Intel is working on a chip, code-named Yamhill, that could read both types of software, although the chip won't come out unless Itanium sales really plummet."

64-bit can read 32-bit, just like 32-bit can read 16-bit. Whether it is the hardware comprehending it or if it is software, any 64-bit system can be made able to read 32-bit. However, for Yamhill to be a bilingual Itanium still remains to be seen.

Besides...
"Otellini said Intel is not planning chips that can handle both 32-bit and 64-bit code. Bilingual chips, such as rival AMD's forthcoming Opteron, could be cheaper and easier for corporations to adopt, analysts have said, because much existing software will work on them."

So it seams like it is just hype to me.

<b>"Sometimes you can't hear me because I'm talking in parenthesis" - Steven Wright</b> :lol: 
June 19, 2002 8:37:08 PM

Umm yeah the naming convention sux. Do you have any suggestions. Maybe they should let an art or english lit major name the processors from now on.

HULK SMASH!!!
June 19, 2002 8:40:26 PM

I don't think they will change the name. It also depend on marketing. Plus they spend lots of money on marketing that name and would want to goto waste. It's all about the Name man. Just like the "Coke" they wouldn't change it even if they could find a better name.

KG

"Artificial intelligence is no match for natural stupidity." - Sarah Chambers
June 19, 2002 8:52:14 PM

You are right, it can read it. But it can't process because it will lack the data that it needs to do so. They way Intel designed is to make you write new code either way it goes.


When I say data I mean info coming through issue ports to the register. That info is neededto determine where to send it from the registry and where to store it in the registry.
HULK SMASH!!!<P ID="edit"><FONT SIZE=-1><EM>Edited by gal128 on 06/19/02 04:57 PM.</EM></FONT></P>
June 19, 2002 9:10:16 PM

Quote:
<i>gal128 says:</i>
The wrench in their plan will be if AMD's Opteron can perform at or above McKinley levels

I'm willing to put money on Opteron not outperforming McKinley. Opteron will be a good processor for other reasons, but I don't think it has a chance against McKinley in pure 64-bit performance.

Quote:
<i>gal128 says:</i>
I believe Intel trying to force people towards 64bit programming is actually a good thing but I think they are going about it in the wrong way.

I agree on both counts, but I don't think they or the people buying Itanium right now care about 32-bit. Would they sell more if Itanium was "bilingual"? Probably, but I don't think they're too concerned about it at this point in time. When they start pushing IA64 for desktop, I would think they'll have some sort of plan for 32-bit.

Quote:
<i>Bront says:</i>
Are we out of good CPU names?

Judging by the names AMD has trademarked, it appears so. But brand/name recognition is important, Intel knows that better than anyone else in the business. Compare the amount of times you've seen "Athalon" typed out here versus how many times you've seen "Pentnium".

Quote:
<i>Bum_JCRules says:</i>
64-bit can read 32-bit, just like 32-bit can read 16-bit. Whether it is the hardware comprehending it or if it is software, any 64-bit system can be made able to read 32-bit.


Of course, but the issue is speed. I heard from a semi-reliable source (personal, not online) that Intel was changing the way Itanium handled 64-bit with McKinley, that it would be much faster. I haven't heard anything since then however, so I'm starting to doubt the accuracy of the information.

<font color=blue>Hi mom!</font color=blue>
June 19, 2002 9:18:36 PM

While I agree on the marketing point, eventually they will have to change the name. The only reason the original Pentium wasnt officially called a 586 was because they could not get copyrights or trademarks to that name. And like my sarcastic example from above a Pentium XXXV would be pretty absurd...plus eventually Intel will completly abandon the x86 architecture. So having a processor name based on a processor that it no longer has anything in common with is also absurd...
just my .02.
June 20, 2002 2:03:39 PM

"Of course, but the issue is speed. I heard from a semi-reliable source (personal, not online) that Intel was changing the way Itanium handled 64-bit with McKinley, that it would be much faster. I haven't heard anything since then however, so I'm starting to doubt the accuracy of the information."

I totally agree. I was just bringing up the fact that the article could be a bunch of bologna.

As for your source, for competition's sake I hope that they do have something to help the 32-bit application problem.

<b>"Sometimes you can't hear me because I'm talking in parenthesis" - Steven Wright</b> :lol: 
June 20, 2002 5:25:00 PM

Quote:
64-bit can read 32-bit, just like 32-bit can read 16-bit. Whether it is the hardware comprehending it or if it is software, any 64-bit system can be made able to read 32-bit. However, for Yamhill to be a bilingual Itanium still remains to be seen.

That statement has about as much credability as saying that a PC can run a Macintosh's software because they're both 32-bits.

If they are the same processor architecture, then sure a 64-bit could run a 32-bit. However the whole point of Itanium is that it is <i>not</i> just another x86 processor extended to 64-bit operation. The Itanium is barely even related to the x86 architecture. This is why software for the Itanium requires a re-write, because it is a whole new architecture. Further, this is also why trying to execute x86 instructions on the Itanium is so slow, because the Itanium doesn't run x86 natively. To run x86 code, it has to <i>emulate</i> it. Intel built x86 emulation into the Itanium, probably mostly for marketting purposes. (Since no one in their right mind would use an Itanium for 32-bit x86 software anyway.) Hence, such crappy speeds for 32-bit software on the Itanium.

Sure, <i>in theory</i> any processor could emulate any other processor with enough emulation software thrown at it. However, as anyone who consistantly uses (or writes) emulators knows, <i>in practice</i> they're hardly ever 100.00% compatible and the further each the processor's architecture is from the other, the slower the emulation can get.


Tech support said take a screen shot.
Putting it down with my .22 was the humane thing to do.
June 20, 2002 5:42:12 PM

"That statement has about as much credability as saying that a PC can run a Macintosh's software because they're both 32-bits."

hehehe :lol:  LOL

That was one of my points. I was making fun of the article. It is just some hype article.

I said that...

"any 64-bit system can be made able to read 32-bit."

This is correct. The performance will not be anything to write home to mother about but it will work.

I'll give you an example. A friend of mine is working on a project for NASA. They are using a cluster running one OS to emulate system running a different OS. That emulated system is then emulating another system which is running a third OS. Messed up and slow, but hey if someone wants to do and pay for it, God bless them. (I love how our governmemt wastes money.)

Anyway, I know what you are saying. I was only bad-mouthing the article.

<b>"Sometimes you can't hear me because I'm talking in parenthesis" - Steven Wright</b> :lol: 
June 20, 2002 6:26:01 PM

I agree. I'm also willing to be that in pure 64-bit performance, Opteron won't be able to match McKinley.

And Bront, CPU names don't really matter. You could come up with new names, but Intel probably won't (it's a marketing thing). For example, they want all the CPU's in the Pentium family to be named "Pentium [number whatever]"

Quote:
<i>originally written by Bum_JCRules</i>
As for your source, for competition's sake I hope that they do have something to help the 32-bit application problem.

You do realize that Itaniums ARE NOT SUPPOSED TO RUN IN 32-BIT! I have said this countless times to people and they nvever listen. Itaniums are pure 64-bit CPU's. Period. Intel threw in 32-bit emulation for no real reason IMHO.




<P ID="edit"><FONT SIZE=-1><EM>Edited by Dark_Archonis on 06/20/02 02:28 PM.</EM></FONT></P>
June 20, 2002 6:54:34 PM

Quote:
You do realize that Itaniums ARE NOT SUPPOSED TO RUN IN 32-BIT! I have said this countless times to people and they nvever listen. Itaniums are pure 64-bit CPU's. Period. Intel threw in 32-bit emulation for no real reason IMHO.


I agree. 32-bit would be nice, and would probably broaden their market. But they don't have an absolute need to include that until it hits the mid-server down to desktop market.

<font color=blue>Hi mom!</font color=blue>
June 20, 2002 7:16:31 PM

However we've seen some companies who would want to transition their 32-bit servers to 64-bit, I think THAT can be a major disadvantage to those who do that, which is why it's always a good thing to ensure backwards compatibility. It's a bit like PS2 over PSX, or moving from Win98 to WinXP. You don't change the entire file system and everything so nobody can get into an NT world!

--
:smile: Intel and AMD sitting under a tree, P-R-O-C-E-S-S-I-N-G! :smile:
June 20, 2002 7:46:50 PM

Companies that are interested in 32-bit code probably wouldn't be interested in Itanium anyway. Why would you have a high-end server running Windows 64-bit and use it with your old 32-bit code?
32-bit will be important later, when Itanium filters down to the lower markets. Not now.

<font color=blue>Hi mom!</font color=blue>
June 20, 2002 8:06:12 PM

Quote:
That was one of my points. I was making fun of the article. It is just some hype article.

It was pretty laughable if you take the author's point of view, but at least it shows that Intel is serious about Itanium.

Still, I'm kind of scared to think that Intel is <i>still</i> on .18 micron for their Itaniums. You would think they'd have moved to .13 micron by now. **shrug** Oh well. Some day...

Quote:
I said that...

"any 64-bit system can be made able to read 32-bit."

This is correct. The performance will not be anything to write home to mother about but it will work.

Yep. Gotta love emulation. It keeps the world going. I don't know what I'd do without my C=64 emulators. **shrug** I'd get bored I guess.

At least I can ditch my crappy Mac emulator now that I've got an actual Mac laying around. (I picked it up for a buck at a silent auction.) That's right folks, they're practically giving Macs away these days. :-p

Now I just have to figure out why in the hell my monitor won't work right on the Mac. It just goes all wonky and keeps scrolling by at high speed. I just don't get it. Do Macs have funky refresh rates or something? Oh well... I'll find a decent way to play RoboWar one of these days.

Quote:
I'll give you an example. A friend of mine is working on a project for NASA. They are using a cluster running one OS to emulate system running a different OS. That emulated system is then emulating another system which is running a third OS. Messed up and slow, but hey if someone wants to do and pay for it, God bless them. (I love how our governmemt wastes money.)

Oh geeze. I can see it though. I've been there. Thank goodness I got out. NASA sounds to be hardly any different than the military. There's the right way, the wrong way, and the military way. Sure, the military way will work (at least usually), but it sure isn't the best way of doing it.


Tech support said take a screen shot.
Putting it down with my .22 was the humane thing to do.
June 20, 2002 8:15:45 PM

mmmm so windows 64 bit (IA-64) is already out then?

and Windows 64 bit (x86-64) is being developed?

what are the advantages of one system over another (or am i opening up a whole can of worms and trolling here :) 

will IA-64 eventually replace all x-86 ?

<font color=red>FUDweiser......True</font color=red>
June 20, 2002 9:04:15 PM

Quote:
mmmm so windows 64 bit (IA-64) is already out then?


Officially, no. In reality, yes.

Quote:
and Windows 64 bit (x86-64) is being developed?


Officially, no. In reality, yes.

Quote:
what are the advantages of one system over another (or am i opening up a whole can of worms and trolling here :) 


Opening up a can of worms, yes. Trolling, not in the least. I've posted a couple of threads on this, but not many people made serious input. I'm not sure whether that's because most people haven't looked into it to learn about both, or whether they're just not interested. No big deal.

Quote:
will IA-64 eventually replace all x-86 ?


I assume you mean x86 <i>and</i> x86-64. Intel hopes it will, but we'll have to wait and see.

<font color=blue>Hi mom!</font color=blue>
June 20, 2002 10:21:02 PM

That's why I said those that are transitioning. Some companies start small and therefore they find no big need for more RAM and whatever quirks 32-bit may bring up later on, and thus want to change to the bigger world. They want to transition but smoothly and not hardly like Intel did. THAT IMO is a point to be taken for why a 64-bit only with possible 32-bit emulation, is not the best. You can't defend a one-feature CPU, when it could be nothing but advantageous and helpful to have 2!

--
:smile: Intel and AMD sitting under a tree, P-R-O-C-E-S-S-I-N-G! :smile:
June 20, 2002 10:31:37 PM

Quote:
Some companies start small and therefore they find no big need for more RAM and whatever quirks 32-bit may bring up later on, and thus want to change to the bigger world.


I can think of no real-world examples of why a company would want to use an Itanium as a transition processor, whether it had full 32-bit peformance or not.

Quote:
You can't defend a one-feature CPU, when it could be nothing but advantageous and helpful to have 2!


Of course it would be. But it doesn't matter in the market the Itanium is currently targeted at. Last I checked, people using huge Itanium machines to simulate the effects of nuclear weapons on the environment with specially written software didn't care about running under Windows 95.

<font color=blue>Hi mom!</font color=blue>
June 20, 2002 10:34:11 PM

I understand that, I was more pointing on the overall of the 64-bit market. Intel can't let Itanium stay niche you know!

--
:smile: Intel and AMD sitting under a tree, P-R-O-C-E-S-S-I-N-G! :smile:
June 20, 2002 10:36:44 PM

They don't plan on it, but for now they're ok with it.

<font color=blue>Hi mom!</font color=blue>
June 21, 2002 6:01:34 PM

The only thing I'll say on this subject is that Intel has said over and over again that they WILL NOT build a bilingual chip. Period. If they do, it will be a true triumph for AMD, showing their insight was better than the mighty Intel. I personally believe that Intel will just push their current chips to the limit, add more cache, then move to a .09 SOI process. At the same time, they'll continue making "Itanics" until they can get the computing community to decide that it's time to move past the x86 ISA and into IA-64. I don't really think that they care if AMD takes care of some of the transition. Is that smart? Time will tell.

She said "I love a man in tight jeans" and I said "They're not supposed to be tight I just got fat."
June 21, 2002 6:06:47 PM

Quote:
Intel has said over and over again that they WILL NOT build a bilingual chip.


Yes, I posted that above.

Quote:
At the same time, they'll continue making "Itanics" until they can get the computing community to decide that it's time to move past the x86 ISA and into IA-64. I don't really think that they care if AMD takes care of some of the transition.


I'm not so sure about this, but they have sunk a lot of money into the development of IA64, and I know they aren't going to give it up for no reason.

<font color=blue>Hi mom!</font color=blue>
June 21, 2002 6:21:41 PM

Quote:
I personally believe that Intel will just push their current chips to the limit, add more cache, then move to a .09 SOI process.

The day Intel uses SOI is the day that hell freezes over and the Olsen twins become become real actresses. Proof is <A HREF="http://www.siliconstrategies.com/story/OEG20020617S0059" target="_new">here</A>

Besides that, carry on you Intanium discussion. :smile:
June 21, 2002 8:08:14 PM

The way I see it, Intel are going to hold the upper hand. If X86-64 takes off, then Intel could just bolt Yamhill onto the Itanium (don't ask me how though...), then they'd have a processor which can take care of both 64-bit architectures. That could yet be the true purpose of Yamhill (insert sinister music here).
June 21, 2002 8:57:49 PM

That's an interesting thought, but it would be a large (physically and logically), expensive processor that's difficult to design, produce, etc.

IA64 and x86-64 are completely different, I would assume they could be integrated, but it would basically be two CPUs in one, perhaps sharing a large cache.

They are more likely IMO to add x86-64 functionality to Xeon, as hinted at in the third quote in my first post.

<font color=blue>Hi mom!</font color=blue>
June 21, 2002 11:08:32 PM

Well, one question is whether the x86-64 thing will even become popular. If it does, then Its not necissarily a problem for Intel. Intel can:

A. Say who cares, for truly fast preformance.. go with McKinley

B. Put out a faster version of x86-64 and say AMD was first but we are faster (which is what really counts)

c. Blow off the low-end server market and concentrate of high end server, and high end desktop. If they corner those two markets... it would likely make up for losing the low-end server market.

Benchmarks are like sex, everybody loves doing it, everybody thinks they are good at it.
June 21, 2002 11:16:51 PM

A and C seem sort of the same to me, am I missing a difference between them?

<font color=blue>Hi mom!</font color=blue>
June 21, 2002 11:58:01 PM

Quote:
Well, one question is whether the x86-64 thing will even become popular.


That's what I was thinking. Until now AMD has had privilege to take advantage of Intel’s effort of enabling new instruction set. For example SSE and SSE2. When Hammer releases later this year with new instruction set (x86-64), who will be supporting this? I hear Microsoft will be releasing a version of Windows with this support but I don’t hear the release date. I have yet to hear any other companies that will be releasing application that can take advantage of x86-64. One of the advantages Intel has is that they have their own team who writes compiler for their processors. Would we except AMD to produce compiler for x86-64 of should we wait for Microsoft to integrate the support in their DevStudios product. We all know how good the Microsoft compilers are.

Anyways there is a long way before x86-64 can be successful.

KG

"Artificial intelligence is no match for natural stupidity." - Sarah Chambers
June 22, 2002 4:31:58 AM

Your right enthusiast, AMD's X86-64 has a long way to go before success. X86-64 has just as much chance as Intel's IA64, however. The difference is that Intel has had a longer go at it, and more capital to push it.

My question is where is the industry hoopla for IA64? Computex this year showed industry wide hardware support for Hammer, from VIA to ATI. Software support has been demonstrated with Windows and Linux with much fanfare. Where is that for IA64?

My prediction is that IBM will eventually eat away at the industry server market with the XSeries S390 systems. Leaving Intel and AMD fighting for market in the affordable server end. Sun has no idea what they want to do. In the end what is left is the consumer market. In that realm I predict X86-64 has much more chance for success then IA64. The reason I give to support this is that X86-64 is bilingual, allowing support for my favorite games. In addition X86-64 will be powerful "enough" to take consumer computing to the next level.

On top of all this X86-64 will be affordable. There is no better influence then price for adoption of "NEW" technology.

Let me know how you feel.
June 22, 2002 5:18:55 AM

I dont think 64-bit will be remotely usefull to the consumer market for a at least a few years. Who needs 2 gigs of RAM?
The x86-64 uses at the consumer have been debated ad-nausium. so i will stop there.

Benchmarks are like sex, everybody loves doing it, everybody thinks they are good at it.
June 22, 2002 6:04:43 AM

I remember people who had 16 megs of ram and seeing people who had 32 and 64 megs of ram (why do you need that much) Or some people saying why do you need 128 megs. As speed gose up so will the ram. I m hitting 1.53 gigs of ram. Hay I was smart when ram prices where cheap. So I got alot of ram.
June 22, 2002 11:08:34 PM

The analogy you said is dead on right and I agree, however 4Gigs of RAM will be a while before it is thoroughly needed. WinXP needs 512MB as a sweet spot to compute well, the Longhorn 3d Windows will require much more I guess, but a huge sum of mem from the video card especially. However the analogy of things going fast requiring more, is completly right.

--
:smile: Intel and AMD sitting under a tree, P-R-O-C-E-S-S-I-N-G! :smile:
June 23, 2002 2:33:48 AM

It probably won't hit the low end for a few years, but then again, with the rate that they're going to release Itanium "sig"cessors it won't be long.

Note: there may or may not be an intended pun in the previous paragraph

And also, did you know, that for the 32-bit emulation, Intel actually put in a Pentium core equivalent for the emulation? Hence, the slow speeds in 32-bit. I heard rumours that for McKinley, they beefed up that emulation core.

And Eden, are there actually any real world examples to these companies who want 32/64 CPU's?

FatBurger, you're right that Itanium will get rid of it's "niche" because Intel is going to aggressively push IA-64 for at least the next three years. And who knows what they're gonna do for the P5?

Texas, you're also right. For the consumer, 64-bit is not needed right now, but remember, IA-64 is NOT aimed at average joes right now.

----------------------------------------------------------

Montecito & Chivano; Intel's Big Guns.
June 23, 2002 10:02:43 PM

The X86 emulation in the Itanium is handled by converting the X86 code into IA-64 code, processing it, and then converting it back into X86 code. The reason why Itanium is so slow at X86 operations is because converting CISC code into VLIW/EPIC code simply cannot be done efficiently. Unless Intel changed the emulation system in Itanium 2, I don't expect too much of an improvement.
June 23, 2002 11:44:27 PM

I reckon we had a user once here, many months ago looking into a cheap 64-bit solution with fairly decent 32-bit so they can transition. It's a long time ago so I have vague memory on this one. But back then I could not suggest waiting for Opteron, and Itanium was the closest one, yet it sucked at transitions.

--
:smile: Intel and AMD sitting under a tree, P-R-O-C-E-S-S-I-N-G! :smile:
June 24, 2002 1:05:57 AM

Also, Intel doesn't think extending inferior instruction set (x86) is a good idea for 64bit. They wanted to get away completely from x86 so they design VLIW/EPIC. If Intel wanted to go AMD way and save lots of money they could have done it easily.

KG

"Artificial intelligence is no match for natural stupidity." - Sarah Chambers
June 24, 2002 5:16:44 PM

Quote:
I reckon we had a user once here, many months ago looking into a cheap 64-bit solution with fairly decent 32-bit so they can transition.


I remember that, and there was no concrete reason he wanted 64-bit. He simply thought it would look good on the budget proposal, most likely.

<font color=blue>Hi mom!</font color=blue>
June 24, 2002 7:19:42 PM

Still, to me it proves that those small rookies companies will eventually WANT to go to something much bigger. I won't rest my case about the fact some do need the transition even in the market sector.

--
:smile: Intel and AMD sitting under a tree, P-R-O-C-E-S-S-I-N-G! :smile:
!