Gulftown is "harvested" Westmere-EX!

[Nedjo]

We all know how Intel downplayed x64 extension to x86, and how eventually they've absorb them to...

We all know how Intel downplayed "native" multicore design, and how eventually they used it to...

We all know how Intel downplayed harvesting method in multicore binning, and now they've reviled it'll be used in upcoming products:

http://arstechnica.com/hardware/news...e-recovery.ars

"The new, "Nehalem"-based Xeon processor that Intel detailed at ISSCC today has eight cores, or at least it does most of the time. Intel reps described for conference attendees how the chipmaker takes a broken eight-core part and turns it into a working six-core part."

of course they have funky name for it: "cache and core recovery"



"So for instance, if testing and validation finds a defect in a cache slice on a chip, then Intel can disable that slice and sell the chip with lower cache. And likewise with cores, so that you might buy a six-core chip from Intel that was originally produced as an 8-core Xeon but had two defective cores.

The company claims that it can effectively isolate the nonfunctioning cache and cores, so that these extra parts don't increase the chip's power draw by letting through leakage current."

I guess what goes around, comes around

http://www.itwire.com/content/view/14512/1154/

[[XC] gomeler]

Just like the Phenom X3s, I see nothing wrong with this. Hopefully this will mean cheaper 6-core parts as I imagine the 8-core chips are going to cost an arm and a leg. I sure would like two of them though

[freeloader]

Environmental savings!

Why waste all the chemicals that went into the CPUs production when you can simply disable a few bad parts. It works for AMD.

[Shintai]

Uhmm...no

Reason 1: EX is FB-DIMMs only.
Reason 2. EX is quadchannel.
Reason 3. EX is not even pincompatible.
Reason 4. EX got 4 QPI links and not 1.
Reason 5. Gulftown got its own mask.
Reason 6. The first EX is 45nm and not 32nm.

EPIC FAIL!

[onewingedangel]

Considering there will not be a native 6 or indeed 4 core Nehalem based Xeon MP, harvesting a lower cost 6 core salvage bin makes perfect sense.

All six core variants will be salvage cores, so it's not a case of selling both 'good' native 6 cores and 'bad' salvaged cores with higher power consumption.

Edit: The title of this thread is just wrong.

[Shintai]

Only thing you can harvest Nehalem-EX to is Nehalem-EX. 6 cores for the MP platform and not 8 or 12MB cache instead of 16MB cache. No way it can go down to either desktop or Xeon DP.

Westmere-EX is a late 2010 product btw. And it wont just be 8 cores...

[Nedjo]

Disapprove my reasons and we talk. EX is 45nm as one of the biggest "issues".

Until then you troll.

read again title: "Westmere-EX"

"-EX" is to point out to the fact that Westmere-EX is replacement for Nehalem-EX

oh yeah... pinout and IMC doen't to do with the core arch!

Xeons and Cores are the same cores with different package and verification procedure...

[Nedjo]

Westmere-EX is a late 2010 product btw. And it wont just be 8 cores...

true... it'll be 6 cores to!

[onewingedangel]

oh yeah... pinout and IMC doen't to do with the core arch!

Xeons and Cores are the same cores with different package and verification procedure...

For the 3 and 5 series xeons yes, but the 7 series (xeon mp) features a quad channel FB-dimm IMC.

The upcoming Xeon MP's do not use the same dies as the Xeon DP or uni-socket Xeon.

[Nedjo]

The upcoming Xeon MP's do not use the same dies as the Xeon DP or uni-socket Xeon.

true.

this is Xeon UP\DP die:



Same as bloomfield...

And this is Xeon-MP die:

[Shintai]

true... it'll be 6 cores to!

Nope. Thats the EP model.

read again title: "Westmere-EX"

"-EX" is to point out to the fact that Westmere-EX is replacement for Nehalem-EX

oh yeah... pinout and IMC doen't to do with the core arch!

Xeons and Cores are the same cores with different package and verification procedure...

Westmere-EX got nothing to do with Gulftown. They are just as compatible as Lynnfield and Bloomfield. Let me guess, Lynnfield is just salvaged Bloomfields with 1 dead memory channel or dead QPI?

There is no way you can cross the 3 families.
Xeon MP with Nehalem-EX and Westmere-EX is one.
Xeon DP with i7 and gulftown is the other.
i5, i3, mobile etc is the last.

Cores are the same, everything else aint. And no way you can mix it.

Also Westmere-EX is about 9months after the Gulftown release....and your dieshot is a Nehalem-EX on 45nm with 24MB cache, 4 QPI links and 4 FB-DIMM channels. FB-DIMMs got a serial interface. DDR3 got a parallel. Good luck in mixing those too.

[HKPolice]

Ya, Westmere i7 Desktop 6core will be its own die, failed chips will get binned into i5 quad core westmeres.

[Nedjo]

Shintai why are you so ignorant?

Lets start from the begining:

here's Nehalem core



By adequate pinout connection you can enable\disable mem. channels... hell if you don't install three DIMMs only two channels of IMC works... same goes for QPI... Nehalem has four of them, but in Bloomfield only one works... so yes Lynnfield is a little bit disabled Bloomfield... but look at that as a feature that will make complete platform more affordable than Bloomfield platform!

[onewingedangel]

Sorry, but you're plainly wrong.

Bloomfield has QPI, lynnfield has DMI.
Bloomfield has no on board PCIe controller.

How exactly could you cripple a Bloomfield to a Lynnfield again

[Hornet331]

i fail to see how this articel tells that future westmeres are broken westmere-EX cores?

The articel only tells that intel might harvest 6core variants of the 8core nehalem_ex, so this might mean we see 6core xeon MP cpus.

[Shintai]

Shintai why are you so ignorant?

Lets start from the begining:

No let me start from the beginning since you dont understand anything.

Nehalem-EX and Westmere-EX got SERIALIZED FB-DIMM memory channels(4 of them aswell). You can not use PARALIZED DDR3 there. And the EX CPUs dont hold 3 spare DDR3 channels to use in this case. Not even to talk about the FB-DIMM channels can only talk with an AMB chip. Not DDR3.

Its like trying to plug a PCI card into a PCIe. Not a great success is it? Another issue is that Westmere-EX got more than 8 cores(12 or 16). Only Nehalem-EX got 8 cores. Nehalem-EX is 45nm. That doesnt fit in a 32nm desktop CPU does it? Do you think the EX CPus got the same pins and power input as i7?

Why the other parts doesnt work are already mentioned. But you could ofcourse just point out the DMI interface and the PCIe x16 lanes on your bloomfield dieshot

i fail to see how this articel tells that future westmeres are broken westmere-EX cores?

The articel only tells that intel might harvest 6core variants of the 8core nehalem_ex, so this might mean we see 6core xeon MP cpus.

Precisely!

[Solus Corvus]

Whatever product it ends up as, "cache and core recovery" is a good idea.

[Movieman]

put's on the hat;
Gentlemen:
When you toss in words like "ignorant" when argueing against someone you are saying " Jeez, I can't beat your argument so I'll call you names"
It accomplishes nothing and the next time I see someone in this thread use this tactic I will send them on an all expense paid 30 day vacation from XS..
Take the hint, play nice or stay out of the thread.
Thanks for reading.

[Mescalamba]

Seriously, who cares?

Its normal for GPUs and its used for looong time (disabling SP and creating various versions of same core, depending on what is 100% ok and what not). Same goes for frequency of GPU and CPU, if CPU pass lets say 3GHz and is 100% stable, they sell it as 3GHz CPU, when not, they try lowering and sell for example at 2,6GHz.

Simple using of what is working.

Only one thing is, that harvested AMD CPUs (X2/X3) usually eats as much as full X4, so I wonder how Intel wants to really disable unusable part..

This thread is interesting only from point of how much will these CPUs eat.. but we will know, when they come out.

[K404]

Its kinda intersting how Intel made comments about AMDs X3 line and are now doing the same thing themselves. Bet those comments are being buried by Intel now

I think tis the right thing to do though- less waste is good, in so many ways

[gOJDO]

We all know how Intel downplayed x64 extension to x86, and how eventually they've absorb them to...

Actually Intel were 100% right about x64. There was absolutely no benefit and no need of x64 for a desktop PC at the time when AMD released Athlon64!

We all know how Intel downplayed "native" multicore design, and how eventually they used it to...

It wasn't Intel who was downplaying "native" design. It was AMD who were BS-ing about Intel MCM. Let me refresh your memory, since you have a selective amnesia:

We all know how Intel downplayed harvesting method in multicore binning, and now they've reviled it'll be used in upcoming products:

BS. They never downplayed that method, actually they said that they will be using the same method latter when there will be more cores on their processors:

WN: AMD recently announced it will be releasing a tri-core processor. What are your reactions to that? The theory is they're releasing it at least in part because of the small market for quad-core chips and that it's going to take some time for people to want to move up.

Rattner: I wouldn't make that much of it. This is a yield-improvement technique, plain and simple. IBM and Sony with their Cell processor -- they have eight (processors) on that that chip, and they said, well seven is the actually the number and one is a spare, or one is dead. I'm never quite sure whether there's a dead one or not. It's just like memory chips today … there are thousands of spare memory bits that are there (to ensure sufficient) yield.

So, yes, (AMD has) a four-core product. I'm sure when they looked at their yield losses, they said, "Wow, we can offer a three-core version of this if one of those cores are dead or slow or whatever it turns out to be."

In terms of software, there's software for one core and then there's software for multiple cores. It's not like, oh, we have a three-core problem but not a four-core problem.

WN: According to AMD, this is a specific product for a specific time -- for this catch-up period before quad-core chips take off.

Rattner: Either way, this is strictly transient because when we start thinking about tens of cores, then we actually start thinking about how we won't build 32, we'll build 36 and we'll have those spare cores so when cores fail we fail down to the spec, right?

When cores are relatively inexpensive, they represent a few percent of the total area. Yeah, throw in a couple extras and you'll increase your yields. You'll guarantee the customer a minimum performance level and they don't have to worry about their customers coming back and saying: "Gee, this seems to be slowing down. What's going on?"

Intel CTO Talks Architecture and AMD Rivalry

[freeloader]

Its kinda intersting how Intel made comments about AMDs X3 line and are now doing the same thing themselves. Bet those comments are being buried by Intel now

I think tis the right thing to do though- less waste is good, in so many ways

Exactly. So you have to eat your words every now and again; who will ever forget Bill Gates infamous 640K quote.

[Nedjo]

put's on the hat;
Gentlemen:
When you toss in words like "ignorant" when argueing against someone you are saying " Jeez, I can't beat your argument so I'll call you names"

wait a second Movieman!

What arguments have Shintai presented?! All he's been saying is: you don't know sh*t, and flowing that with sacrasm like: "But you could ofcourse just point out the DMI interface and the PCIe x16 lanes on your bloomfield dieshot "

What argument has he present to counter my opinion??


Only argument that's presented here and that's legit is about Intel's harvesting approach!

All the rest is my personal opinion vs. Shintais one!

And now is OK for him telling me you don't know s*it, and being sarcastic () and I'm politely warned not to use word "ignorant" or I'll be out for 30 days!?!

So now Shintai has monopoly on all subjects regarding Intel, and all the other opinions will be slapped with "you don't know sh*t", and one who dares to point out to this ignorant behavior will get 30 days pinkslip? sad!

[Shintai]

Only argument that's presented here and that's legit is about Intel's harvesting approach!

No its not. Lets see the topic again.

Gulftown is "harvested" Westmere-EX!

The article you link doesnt contain Westmere-EX(32nm Xeon MP) or Gulftown (32nm i7). It however contains Nehalem-EX aka. Beckton.

You made it up and didnt know what you talked about. And then continued with your misinformation and wrongdoing in the rest of your posts.

And now you blame others....

Lets go back to the article you linked:

Intel details eight-core Xeon, "cache and core recovery"
The new, "Nehalem"-based Xeon processor that Intel detailed at ISSCC today has eight cores, or at least it does most of the time. Intel reps described for conference attendees how the chipmaker takes a broken eight-core part and turns it into a working six-core part.

So you just changed it all for the lolz. With some added trolling in the start.

Even if your topic was used it had to be:
Nehalem-EX is "harvested" Nehalem-EX!

Faulty 24MB cache 8 core 45nm Nehalem-EX salvaged as something like 6 core, 18MB cache 45nm Nehalem-EX.

Also Intel have salvaged cache parts for ages. And they never said they wouldnt do it on the cores. Instead they said they didnt make faulty cores. Or enough of them so to speak.

[Hornet331]

Exactly. So you have to eat your words every now and again; who will ever forget Bill Gates infamous 640K quote.

it's even funnier that bill gates never said "that" thing about 640k

@Nedjo

no, he hasn't the monopole, but on the otherhand at least he posts facts... (even if his tone is not polite).

You on the otherhand post an articel about intel, where it is discused, that they may use harvesting teqniques for nehalem-EX, and then conclude westmeres are broken westmere-ex.

That makes absolute no sense at all.