Intel Details Itanium "Poulson" Eight-Core Chip

[nyker96]

Get a load of this chip: 50mb cache on die!

http://www.xbitlabs.com/news/cpu/di...t_Core_Chip_for_Mission_Critical_Servers.html

 

[Idontcare]

50MB of on-die cache is rather considerable. Has got to help with those yields to. Die is large, but yields will be high because of cache/core redundancy, and ASP's will be high because of the market segment itself.

 

[VirtualLarry]

So is something like this what runs the WoW servers?

 

[IntelUser2000]

Last I read, game servers run x86 servers like Opterons and Xeons.

 

[sandorski]

I'm no Analyst, but from what I can glean it looks like Intel has at least eeked out some Profit on Itanium. It certainly has been a dismal failure to meet expectations. Ironically x86-64, AMD's creation, has been much more successful, but more than that, x86-64 is Itanium's largest competitor.

 

[Soundmanred]

It will run a Windows 8 silky and sexy smooth.
gl,,,,gb,,

 

[ShadowVVL]

I ran into the itanium mid last year running around on the net and have always wondered what it was for.

 

[Rubycon]

Just in time for Crysis 2.

 

[IntelUser2000]

-8 cores
-16KB L1 I and D, total 32KB per core
-512KB L2 I, 256KB L2 D per core
-32MB L3 split into 8 blocks
-Two 1.5MB Directory Caches

I'm guessing the rest(to fill up 50MB) is something we normally don't discuss.

 

[pm]

For what it's worth, this is the chip that I work on.

Just in time for Crysis 2.

 

[IntelUser2000]

For what it's worth, this is the chip that I work on.

Location: Fort Collins

Who could've guessed.

 

[RocksteadyDotNet]

12 issue?!

 

[Rubycon]

For what it's worth, this is the chip that I work on.

Any chance of seeing 100 megapixel atomic force microscope die shots?

 

[degibson]

12 issue?!

Relax. Statically scheduled VLIW.

 

[RocksteadyDotNet]

Relax. Statically scheduled VLIW.

The guys that write the compilers must be hardcore.

 

[Voo]

It will run a Windows 8 silky and sexy smooth.

Considering that MS already said it won't support Itanium in following releases.. that seems unlikely.

Not that many Itaniums ran MS Server (or Red Hat) anyway, but it still doesn't make the plattform look any more healthy.

The guys that write the compilers must be hardcore.

Oh certainly - those poor bastards aren't to be envied. Although certainly a nice challenge

 

[Idontcare]

For what it's worth, this is the chip that I work on.

It's gotta feel great to see it getting some public exposure :thumbsup:

 

[Idontcare]

Considering that MS already said it won't support Itanium in following releases.. that seems unlikely.

Not that many Itaniums ran MS Server (or Red Hat) anyway, but it still doesn't make the plattform look any more healthy.

x86 did alright for itself without the benefit of windows for a while too

 

[Voo]

x86 did alright for itself without the benefit of windows for a while too

But you can't say that Windows (well MS) wasn't important in getting it where it's now - Wintel and so.

And sure I'm not saying that you need Red Hat or MS support for every processor and considering that the majority of Itaniums are used with HP's OS (umn whatever its name was) anyhow it may not matter much today. But I think it cements Itanium's role as a niche product for the highend server market (with fierce competition not just from the RISC machines but from it's smaller brother aka Xeon as well) that failed completely in the mass market. I just can't think that Intel is content with how the whole experiment turned out (and even if they're finally making some profit with it, they sunk billions in R&D into it; so from a business perspective hardly a wholesale success isn't?)


As an anecdote my university for example is using Itaniums right now for the computing center (nothing especially large, 1k cores or so) but it's pretty certain that for the next upgrade they will go with x64 CPUs and I've heard of similar decisions in other universities as well.
Obviously only pretty weak anecdotal evidence so make with it what you want

 

[sandorski]

But you can't say that Windows (well MS) wasn't important in getting it where it's now - Wintel and so.

And sure I'm not saying that you need Red Hat or MS support for every processor and considering that the majority of Itaniums are used with HP's OS (umn whatever its name was) anyhow it may not matter much today. But I think it cements Itanium's role as a niche product for the highend server market (with fierce competition not just from the RISC machines but from it's smaller brother aka Xeon as well) that failed completely in the mass market. I just can't think that Intel is content with how the whole experiment turned out (and even if they're finally making some profit with it, they sunk billions in R&D into it; so from a business perspective hardly a wholesale success isn't?)


As an anecdote my university for example is using Itaniums right now for the computing center (nothing especially large, 1k cores or so) but it's pretty certain that for the next upgrade they will go with x64 CPUs and I've heard of similar decisions in other universities as well.
Obviously only pretty weak anecdotal evidence so make with it what you want

Ever since release they've Projected Sales in the $10-20billion/yr range and repeated that projection annually. So far they've managed $4 billionish/yr. Still a decent chunk of change, but nowhere near what they wanted from it.

 

[jhu]

cheap used Itanium2 systems can be had on ebay. I've been tempted to buy one just to play around with, except the wife wouldn't be too happy with yet another computer on the premises...

 

[degibson]

Ever since release they've Projected Sales in the $10-20billion/yr range and repeated that projection annually. So far they've managed $4 billionish/yr. Still a decent chunk of change, but nowhere near what they wanted from it.

It's a hard market to break into.
http://en.wikipedia.org/wiki/File:Itanium_Sales_Forecasts_edit.png

 

[tweakboy]

It will run a Windows 8 silky and sexy smooth.
gl,,,,gb,,

LOL thanks manred. gb