32nm quad core Xeons

[cbn]

http://techpowerup.com/110567/Intel_Readies_13_Westmere-based_32_nm_Xeon_Processors.html

These should fit on LGA 1366 consumer boards right? If so X58 customers might have some affordable choices for 32nm quad core soon.

I'm sure the extra L3 cache (over Core i3) will help at higher overclocked speeds as well.

 

[Idontcare]

nm

 

[jones377]

Some people were very critical about what they called AMD's cripple cores. I wonder what they say about this?

 

[Idontcare]

Savvy product diversification for an artificially segmented market as a means to enrich the ASP topology and maximize shareholder equity?

I hear it is going to be a paradigm shift with synergistic effects between Intel accounting and Otellini's direct-deposit account.

 

[exar333]

Some people were very critical about what they called AMD's cripple cores. I wonder what they say about this?

That it stomps all other quads in terms of performance?

 

[jones377]

That it stomps all other quads in terms of performance?

Anything but owning up to their hypocrisy I guess

 

[Idontcare]

Anything but owning up to their hypocrisy I guess

What did Intel state about AMD's core-harvested chips? I missed it.

 

[jones377]

What did Intel state about AMD's core-harvested chips? I missed it.

I was talking about their followers

 

[soccerballtux]

Savvy product diversification for an artificially segmented market as a means to enrich the ASP topology and maximize shareholder equity?

I hear it is going to be a paradigm shift with synergistic effects between Intel accounting and Otellini's direct-deposit account.

sellout.
so that's what you left TI for...psht.

 

[aigomorla]

the 32nm as i said were 2x qpi chips meant for dual qpi boards.

Or board with 2 sockets.

However the dual socket boards are also the same pin set LGA1366, and they work in X58 boards.

However you will pay a premium for that second qpi which u will never use.

 

[Idontcare]

I was talking about their followers

Got it. Yeah, same thing about MCM versus monolithic in the other camp. Not everybody says it or thinks it though, for either subject. Just business.

 

[Compddd]

What do you think the prices will be for the X5650 & X5620? Hope those are the best bang for the buck like the 920

 

[jones377]

Got it. Yeah, same thing about MCM versus monolithic in the other camp. Not everybody says it or thinks it though, for either subject. Just business.

Yes it's equally bad on both sides. I didn't mean to derail this thread though, and I actually think these chips are a GOOD thing. I always suspected there would be quads on 32nm from Intel by using the hexacores and this confirms it. All we need now are some on the consumer desktop side too.

 

[exar333]

Yes it's equally bad on both sides. I didn't mean to derail this thread though, and I actually think these chips are a GOOD thing. I always suspected there would be quads on 32nm from Intel by using the hexacores and this confirms it. All we need now are some on the consumer desktop side too.

Intel has done similar things with disabling cache before, but I don't recall them selling CPUs with cores disabled. It completely makes sense, and only helps the chip-maker sell more products, which ultimately reduces costs for consumers (ideally).

If these quads are <$600, I will definitely get one. Although I would love a hexa-core, these might be almost as good. Instead of 6-cores at 3.2-3.6ghz (OCd) I could potentially have 4 cores with more cache per core (3MB instead of 2MB) and maybe get 4.0-4.5+ on the OC.

1366 looks like a winner!

 

[aigomorla]

What do you think the prices will be for the X5650 & X5620? Hope those are the best bang for the buck like the 920

Once again you will pay a premium for the second QPI.
Even tho you will never use it.

So 920? No way.

940/950/960 prices? Very likely.

 

[aigomorla]

well seeing how Victor no longer cares anymore, i think i will post this picture here.

He came to a friend of mine going why is this gulftown only 4 cores.
LOL.. my friend scratched his head, and asked me, 32nm quadcore?
Then i went to my sponsor with this picture and said : "Please.. with both my hands open"

It was a few days later victor posted the cpu on the thread asking for help.
He got these guys thinking they were hexcores and not quadcores.

This is how I personally found out 32nm quads were rolling in, but ONLY 2x QPI, with 1xQPI to come Much much later.

The thread in which he is asking for help.
http://www.xtremesystems.org/forums/showthread.php?t=240252

Screenshot of his CPU.

You can thank victor for posting it in a public location.
They were supposed to be HUSH HUSH.

Anyhow.. i expect starting price... roughly 500+ dollars... per lot 1k.
For a 3.2ghz version which is about the same as the 960, expect more then 960 price.

 

[hans007]

Intel has done similar things with disabling cache before, but I don't recall them selling CPUs with cores disabled. It completely makes sense, and only helps the chip-maker sell more products, which ultimately reduces costs for consumers (ideally).

If these quads are <$600, I will definitely get one. Although I would love a hexa-core, these might be almost as good. Instead of 6-cores at 3.2-3.6ghz (OCd) I could potentially have 4 cores with more cache per core (3MB instead of 2MB) and maybe get 4.0-4.5+ on the OC.

1366 looks like a winner!

All the single core celerons are basically disabled core Allendales, or on laptops are disabled core 45nm chips.

The thing with crippling say VT or Hyperthreading or x64 (atom chips in nettops have it, atoms in netbooks dont) is those parts always are working on the chips.

Disabling bad cache or a core, seems ok to harvest a die that can still be sold at least in my opinion. I do think that a lot of the time the intel cache / core is probably ok too, they just want more product differentiation (just like a lot of AMD chips have working cores that are disabled but at least they provide you a way to try to enable it at your own risk)

but disabling a feature that is inherently part of the architecture seems pretty consumer unfriendly.

 

[Idontcare]

but disabling a feature that is inherently part of the architecture seems pretty consumer unfriendly.

Well they (the manufacturer) has two choices when they want to create product differentiation thru feature-set differences.

They can make more of a given cpu that has already been designed and taped out (higher volume, leading to lower cost) and intentionally disable a certain percentage of them.

Or they can go to the effort and expense of creating a new design sans the features that aren't going to be used as selling points in the specific SKU and then manufacture that new taped out design in much lower volumes as they can only use it for sales in that particular market segment...all of which leads to higher costs which would inevitably be passed along to the consumer in the form of higher prices so that the targeted gross margins are maintained.

So as a consumer would you rather pay a higher price for a product that was designed to not contain the features you aren't paying to use anyways, or would you like a cheaper price for the same feature-set but knowing that your chip could have had more features (and sold at a higher/different price point altogether)?

I'd say it is rather consumer friendly to have cheaper products.

 

[exar333]

Well they (the manufacturer) has two choices when they want to create product differentiation thru feature-set differences.

They can make more of a given cpu that has already been designed and taped out (higher volume, leading to lower cost) and intentionally disable a certain percentage of them.

Or they can go to the effort and expense of creating a new design sans the features that aren't going to be used as selling points in the specific SKU and then manufacture that new taped out design in much lower volumes as they can only use it for sales in that particular market segment...all of which leads to higher costs which would inevitably be passed along to the consumer in the form of higher prices so that the targeted gross margins are maintained.

So as a consumer would you rather pay a higher price for a product that was designed to not contain the features you aren't paying to use anyways, or would you like a cheaper price for the same feature-set but knowing that your chip could have had more features (and sold at a higher/different price point altogether)?

I'd say it is rather consumer friendly to have cheaper products.

Great post IDK. One other thing to consider is speed-binning. The chip-maker isn't hurting consumers by binning some chips capable of higher speeds and selling them at lower speeds. When manufacturors sit down and decide on feature sets, they have to weigh production issues AND marketable features. They may have a product series with xMB cache fall into one price range, and a different product series with more cache, fall into another (potentially faulty or purposefully disabled). Features like HT are going to be less "yield-related" and more decided on in terms of if the feature should be present or not.

Selling chips isn't just about clockspeed, its also about features. For example, here may be more incentive to step up to a 2.0ghz CPU w/ VT compared to a 1.86ghz CPU without VT. This product definition allows lower-priced products, higher priced products, and anywhere in-between.

 

[ViviTheMage]

Intel headlights? haha

 

[VirtualLarry]

but disabling a feature that is inherently part of the architecture seems pretty consumer unfriendly.

Intel has ALWAYS been "Consumer unfriendly".

Edit: Just thinking of how "UN-green" Intel products are, because they disable SpeedStep on Celerons. Millions of laptops in use sucking down and wasting more power than they need, shortening battery life and inconveniencing people unneccesarily, just to prop up Intel's profit margins.

AMD really deserves a gold star here, because ALL of their chips, from the top to the bottom of their line-up, have power-saving features.

 

[cbn]

Its Xeon, not Xenon.

Thanks.

I made the corrections in the opening post.

 

[cbn]

Anyhow.. i expect starting price... roughly 500+ dollars... per lot 1k.
For a 3.2ghz version which is about the same as the 960, expect more then 960 price.

I was hoping these chips would be cheaper considering desktop hexcore EE @ 3.33 Ghz was supposed to be $999.

These chips seemed like a much lower bin. However, I can understand adding the second QPI would increase costs.

 

[deputc26]

4.61 ghz at 1.344v, this seems representative of what we can expect from 32nm nehalem

 

[JFAMD]

Got it. Yeah, same thing about MCM versus monolithic in the other camp. Not everybody says it or thinks it though, for either subject. Just business.

To be clear though, we didn't say MCM was bad, we said *unconnected* MCM was bad.

There are benefits of MCM that Intel brought up and we agreed with (time to market, yield, etc.) But there was a big downside of an unconnected MCM which was higher bus traffic, latency and system bottlenecks.

Most of the time that I criticized MCM, the sound bite that made it into the publication was simplified to "MCM is bad" and the subtlety of what I had said was lost. We had MCM on the roadmap at the time that we were making the statements, so that was always in the backs of our minds.

Limiting speed bin, core or cache is fine. The speed bin sorting has been going on for years in the industry. However, limiting features like HT, virtualization or turbo generally lead to more customer confusion. People expect to get X and end up with Y. They can't understand why applications perform differently.

I guess if I had to categorize it, I would group things into "obvious" and "not obvious". When you buy a processor in a system, you see the clock speed, cores and cache listed. Those are obvious and easy for people to understand. Some of the other things are less obvious and more difficult for people to understand what they have/don't have. Not an issue as much for consumers as it is in a data center where you have hundreds or thousands of servers with dozens of different configurations.