Hardware Canucks

Hardware Canucks (http://www.hardwarecanucks.com/forum/)
-   Press Releases & Tech News (http://www.hardwarecanucks.com/forum/press-releases-tech-news/)
-   -   Intel Looks to Launch 8-Core Xeon Nehalem-EX (http://www.hardwarecanucks.com/forum/press-releases-tech-news/29625-intel-looks-launch-8-core-xeon-nehalem-ex.html)

FiXT March 8, 2010 11:00 AM

Intel Looks to Launch 8-Core Xeon Nehalem-EX
 
The server market always gets the cool stuff first. Intel is readying to launch its first 8-core processor to combat the already launched AMD 8 and 12 core Opteron units. Named the Nehalem-EX, these octo-core chips are manufactured on the 45nm process, featuring 8 physical cores each support two simultaneous threads through Hyperthreading. It incl... [ Read full article ]

Oversized Rooster March 8, 2010 11:04 AM

Still 45nm?! Booooooo!

_dangtx_ March 8, 2010 12:43 PM

well 32 would be nice, but still, with a sensible power envelope it would make some of us folders quite happy :)

jurassic1024 March 8, 2010 03:55 PM

Boo?
 
Why do you care if the EX, a server CPU, is 45nm? It's Nehalem anyway, which is 45nm... D'oh!!! Westmere is 32nm AND for the desktop. sheesh. Do you only read headlines then jump to the comment button?

Zero82z March 8, 2010 05:41 PM

Quote:

Originally Posted by jurassic1024 (Post 344125)
It's Nehalem anyway, which is 45nm... D'oh!!! Westmere is 32nm AND for the desktop.

Nehalem is the name of the architecture and has no specific process associated with it. Bloomfield is the 45nm equivalent of Westmere.

Oversized Rooster March 8, 2010 06:54 PM

Quote:

Originally Posted by jurassic1024 (Post 344125)
Why do you care if the EX, a server CPU, is 45nm? It's Nehalem anyway, which is 45nm... D'oh!!! Westmere is 32nm AND for the desktop. sheesh. Do you only read headlines then jump to the comment button?

I read the whole article actually. A particular architecture is not necessarily tied to a specific manufacturing lithographic process.

Remember the Pentium 4? The Northwoods were 130nm and the Prescotts were 90nm. They were both the same Netburst architecture.

You're damn right I'm disappointed that new products are being rolled out using the 45nm process. Intel has clearly shown they CAN produce consumer products at 32nm so it's logical for me to feel disappointment when they don't shift existing quad core and new hexa and octa core CPUs to the 32nm process.

32nm = less heat, less power consumption, more overclocking with less voltage

Generic User #2 March 8, 2010 06:55 PM

i'm pretty sure you're wrong about that.

bloomfield is equivalent to gulftown(super high-end)
lynnfield is equivalent to clarkdale
clarksfield is equivalent to arrandale

Sagath March 8, 2010 07:44 PM

Quote:

Originally Posted by Oversized Rooster (Post 344221)
I read the whole article actually. A particular architecture is not necessarily tied to a specific manufacturing lithographic process.

Remember the Pentium 4? The Northwoods were 130nm and the Prescotts were 90nm. They were both the same Netburst architecture.

You're damn right I'm disappointed that new products are being rolled out using the 45nm process. Intel has clearly shown they CAN produce consumer products at 32nm so it's logical for me to feel disappointment when they don't shift existing quad core and new hexa and octa core CPUs to the 32nm process.

32nm = less heat, less power consumption, more overclocking with less voltage

Its not that easy.

I suggest you read this article: AnandTech: The RV870 Story: AMD Showing up to the Fight to get an idea of what die shrinking entails from a design standpoint

The TLDR is this. Switching nodes usually means a complete redesign of the chip. Shrinking components down means rerunning tracing, redesigning circuit pathing, moving components and die areas around, etc. Or, it means designing the chip in the first place with scalability to the next node in mind. Intel didn't bother with it. Why? Because its a lot easier to build a monolithic eight core chip on a mature known process, then hope and pray all 8 cores work on a new process you just started on...

But I'm sure they are hiring guys like you to fix their business model for them. Actually, I heard they just fired all their PhD engineers with 10+ years of schooling, and 5+ years of job experience to open a slot up! :whistle:

Oversized Rooster March 8, 2010 07:52 PM

Quote:

Originally Posted by Sagath (Post 344244)
Its not that easy.

I suggest you read this article: AnandTech: The RV870 Story: AMD Showing up to the Fight to get an idea of what die shrinking entails from a design standpoint

The TLDR is this. Switching nodes usually means a complete redesign of the chip. Shrinking components down means rerunning tracing, redesigning circuit pathing, moving components and die areas around, etc. Or, it means designing the chip in the first place with scalability to the next node in mind. Intel didn't bother with it. Why? Because its a lot easier to build a monolithic eight core chip on a mature known process, then hope and pray all 8 cores work on a new process you just started on...

But I'm sure they are hiring guys like you to fix their business model for them. Actually, I heard they just fired all their PhD engineers with 10+ years of schooling, and 5+ years of job experience to open a slot up! :whistle:

What you described here deals with the details of engineering and product process. What I said was that you could have the same architecture chip on different production processes. Redesigning circuit paths, etc doesn't mean it's a new architecture.

And to be honest, I don't care what the problems are to move to 32nm because I'm an end user. If I was an electrical engineer in that field I might have cared but, it's not my problem. I am not an engineer and all I care about is getting more performance for my dollars spent.

Nor do I have the desire to enter this field of work...so I don't see why you're getting all personal.

We all "wish" for stuff. Sometimes the manufacturers put out impressive new products and sometimes they fall short of consumers' expectations or desires. That's life.

Sagath March 8, 2010 08:11 PM

Quote:

Originally Posted by Oversized Rooster (Post 344247)
What you described here deals with the details of engineering and product process. What I said was that you could have the same architecture chip on different production processes. Redesigning circuit paths, etc doesn't mean it's a new architecture.

And to be honest, I don't care what the problems are to move to 32nm because I'm an end user. If I was an electrical engineer in that field I might have cared but, it's not my problem. I am not an engineer and all I care about is getting more performance for my dollars spent.

Nor do I have the desire to enter this field of work...so I don't see why you're getting all personal.

We all "wish" for stuff. Sometimes the manufacturers put out impressive new products and sometimes they fall short of consumers' expectations or desires. That's life.

You're stomping your feet and shouting "Its not fair!" over something that's not realistic, and you fail to acknowledge it. Yes, you can have the same basic architecture on a new node. However, that doesnt mean they dont have to redesign the chip. Hence why I linked you the article to read.

Wish all you want for the sun, stars and a billion dollars, but the fact remains that a node shrink would entail a redesign for a monolithic 8 core chip on a brand new node. They arnt going to eat those kinds of resources up so you can overclock it 200mhz higher.


All times are GMT -7. The time now is 03:09 AM.