Intel 32nm Westmere Desktop Processor Roadmap Exposed

[jones377]

Originally posted by: Idontcare

Originally posted by: IntelUser2000
Probably not the same thing as seeing in person, but the wafer at the picture below is apparently containing Nehalem-EX.

http://www.heise.de/resale/IDF...Plaene--/zoom/114504/0

The L3 is actually extremely small in comparison to the core in Nehalems. Per MB die size is only 6mm2.

Thanks for the link, I had seen that photo before but for the life of me could not track it down afterwards :beer:

Yeah it's huge, look at it relation to Pat's fingernails.

Makes me wonder just how freaken big pm's toes must be if he thinks those monsters are "about the size of a large toenail"...:laugh: Yeah, maybe Shaquille O'Neal's toenails

I gotta ask you something. These wafers that CEO's like to show off at these trade shows... I assume they failed catastrophically at some point in the manufacturing process right? Or would a wafer survive being displayed like that and then cut up into products?

 

[Idontcare]

Originally posted by: jones377
I gotta ask you something. These wafers that CEO's like to show off at these trade shows... I assume they failed catastrophically at some point in the manufacturing process right? Or would a wafer survive being displayed like that and then cut up into products?

Yeah, those are dead wafers. Usually not even fully processed, meaning you are probably looking at a wafer that may have been thru all the steps of the front-end but died from some processing snafu at M1 or somesuch in the back-end. (just example)

A good wafer would survive being displayed like that but it would take an inordinate amount of effort to recover the wafer, stripping back the topmost processed levels and reworking the wafer (having extra risks in its own right).

Generally in development there is no shortage of dead half-processed wafers for CEO's to parade in press junkets. They do like to reclaim the wafers though as the substrate alone is worth a good $300-$400 just as scrap silicon, so you won't see them being handed out like free 1GB thumbdrives at trade shows.

Originally posted by: Denithor
So no 4/8 32nm chips for LGA1366?

Yep, looks like 1366 goes straight to Gulftown 6c/12t goodness. But knowing Intel the bclck will be upped from current 133MHz to something like 200MHz just to totally screw anyone who doesn't get the extreme edition :laugh: Forever locking the i7 920 into filling the role of Q6600 for the remainder of 32nm's shelf-life. (I joke, I hope )

 

[nitromullet]

Originally posted by: Denithor
So no 4/8 32nm chips for LGA1366?

Because the LGA1156 isn't as interesting due to the reduced PCIe lane count. Do they have SLI/CF enabled for LGA1156? Even at the lower bandwidth most cards should run ok, I suppose.

And that 2/4 32nm, while interesting, just won't have enough horsepower going forward for games. Today's games can max two cores already so there's no advantage from hyperthreading on a dual-core (for gaming).

CF is supported on anything with two physical PCIe 16x slots, even if there are only 8x/8x lanes. I doubt that NVIDIA would qualify 8x/8x boards for SLI though.

 

[Phynaz]

Originally posted by: jones377

Originally posted by: Idontcare

Originally posted by: IntelUser2000
Probably not the same thing as seeing in person, but the wafer at the picture below is apparently containing Nehalem-EX.

http://www.heise.de/resale/IDF...Plaene--/zoom/114504/0

The L3 is actually extremely small in comparison to the core in Nehalems. Per MB die size is only 6mm2.

Thanks for the link, I had seen that photo before but for the life of me could not track it down afterwards :beer:

Yeah it's huge, look at it relation to Pat's fingernails.

Makes me wonder just how freaken big pm's toes must be if he thinks those monsters are "about the size of a large toenail"...:laugh: Yeah, maybe Shaquille O'Neal's toenails

I gotta ask you something. These wafers that CEO's like to show off at these trade shows... I assume they failed catastrophically at some point in the manufacturing process right? Or would a wafer survive being displayed like that and then cut up into products?

Edit: IDC beat me to it.

 

[ilkhan]

Originally posted by: IdontcareYep, looks like 1366 goes straight to Gulftown 6c/12t goodness. But knowing Intel the bclck will be upped from current 133MHz to something like 200MHz just to totally screw anyone who doesn't get the extreme edition :laugh: Forever locking the i7 920 into filling the role of Q6600 for the remainder of 32nm's shelf-life. (I joke, I hope )

I hope not. Unlike the FSB, theres no reason to up the bclock speed, at all. Everything is based on multi's, rather than the actual bus speed.

 

[Idontcare]

Originally posted by: ilkhan

Originally posted by: IdontcareYep, looks like 1366 goes straight to Gulftown 6c/12t goodness. But knowing Intel the bclck will be upped from current 133MHz to something like 200MHz just to totally screw anyone who doesn't get the extreme edition :laugh: Forever locking the i7 920 into filling the role of Q6600 for the remainder of 32nm's shelf-life. (I joke, I hope )

I hope not. Unlike the FSB, theres no reason to up the bclock speed, at all. Everything is based on multi's, rather than the actual bus speed.

On the desktop there was no reason to increase the FSB either once we got above 800MHz.

I don't fault Intel for doing it (raising FSB, or bclock if they do it too) as it wasn't really done to screw enthusiasts (I joke about it, but am never seriously thinking its true)...upping the FSB was done to create the perception of a value-gap that upgrading your mobo and ram was intended to fill. Motherboard and ram makers appreciate when a vendor like Intel does little things that helps put food on everyones table.

 

[PCTC2]

Originally posted by: Idontcare

Originally posted by: angry hampster

Originally posted by: PCTC2

Originally posted by: angry hampster
Seems like the Nahelem early adopters really got dicked on this one. I may be in for one of those new dual core/quad thread Clarkdales.

Uhh. I don't think I"m dicked for one. If you happen to notice, the Extreme segment of the market will remain the X58 and Skt 1366. Sure, Lynnfield with an integrated PCIe controller and IMC looks nice, but I sure like my i7. Plus, Skt 1366 on the X58 chipset gets Gulftown w/ 6 cores and 12 threads on the X58. I'm pretty sure I'll just plop one of those into my X58 and I'm not going to feel dicked. At all. Have fun with your quad-core when I have my sex-core.

Ahh I didn't see that they're keeping 1366 around. Also, I buy what I need, not what gets e-peen cred on geek forums.

epeen is not about what you bought, it is about why you bought it. Be it a 965 or a a lowly Cedar Mill.

On this particular geek forum, knowing what you need and knowing the lowest cost pathway to procuring what you need is what gets you cred of the non-epeen variety. (also being a helpful poster to your colleagues works too )

Not every 965 system out there was procured to satiate an epeen campaign.

You can be epeen about getting a dual-core 4GHz E8600 system if given what you do would have been satisfied with a much cheaper netbook.

:thumbsup: I use my i7 for mainly virtualization. Hyperthreading on a quad-core allows the most efficient density of VMs on my host.

 

[aigomorla]

until i see an i5 on sale @ newegg soon.

and i mean like within a few weeks.

Ima call shins on this entire presentation.

As i said, according to that magic presentation we should have i5's.

Yet where the FAK are they?

Forget the i5, Where the hell is even a LGA1156 preview???

Its called

D E L A Y S.

get used to them this year people. Im seeing them on my sample requests.

 

[Martimus]

Originally posted by: Idontcare

Originally posted by: ilkhan

Originally posted by: IdontcareYep, looks like 1366 goes straight to Gulftown 6c/12t goodness. But knowing Intel the bclck will be upped from current 133MHz to something like 200MHz just to totally screw anyone who doesn't get the extreme edition :laugh: Forever locking the i7 920 into filling the role of Q6600 for the remainder of 32nm's shelf-life. (I joke, I hope )

I hope not. Unlike the FSB, theres no reason to up the bclock speed, at all. Everything is based on multi's, rather than the actual bus speed.

On the desktop there was no reason to increase the FSB either once we got above 800MHz.

I don't fault Intel for doing it (raising FSB, or bclock if they do it too) as it wasn't really done to screw enthusiasts (I joke about it, but am never seriously thinking its true)...upping the FSB was done to create the perception of a value-gap that upgrading your mobo and ram was intended to fill. Motherboard and ram makers appreciate when a vendor like Intel does little things that helps put food on everyones table.

Hey! You stole my conspiracy theory for the bclock increase!

JK, I actually got the idea from you a while back while you made the same conjecture about why FSB increases were happening - it may be increased for political reasons more so than performance reasons.

Although this time it would be harder to sell as an actual performance upgrade as it doesn't even give a theoretical performance advantage like the a FSB increase would. For this reason I see it as less likely now that I have put some actual thought into it (rather than just parroting what you had said about why Intel increased the FSB from 800MHz to 1066MHz and then 1333MHz.)

 

[Kraeoss]

indeed this is the year of DELAYS lol recession causes delay salary amount lol

 

[aigomorla]

as i said.

we havent even seen a LGA1156 board preview yet.

I havent seen one in real life. Infact the only one ive seen is on the forums like this:

http://www.pcgameshardware.com...ets-LGA-1155-and-1156/

I dont know WTF they even look like, and i have no idea how much they cost.

Usually we know all these info when there about to come out.

Dude... if thats all were getting now, it means about 3-4 months b4 we start seeing stuff for it.

So no im calling shins on this entire article, and also saying Intel picked up a BAD HABIT from AMD.

everyone say it with me...

S H I N S...

32nm will not come out b4 45nm and 45's arent even on the market yet. So...

S H I N S!!!!

Shiet.. my gainestown in my sig arent even cleared for NDA yet. And IDC you know how long ive had these girls.

 

[ilkhan]

What?
45nm has been out and in retail for a year
bloomfield has been out for almost a quarter already
lynnfield/clarksfield will be out for back to school season
arrandale/clarksdale will be out for Christmas season

Quit being quite so pessimistic. Theres nothing new for you to compare, it'll be the same case and cooling design as we've had with bloomfield.

 

[VirtualLarry]

Will 1366 aftermarket heatsinks fit on 1156 chips? That's my big question. If the answer is no, then this will wreak havor on the heatsink community, if they have to design clips for yet another socket.

 

[aigomorla]

Originally posted by: ilkhan
What?
45nm has been out and in retail for a year
bloomfield has been out for almost a quarter already
lynnfield/clarksfield will be out for back to school season
arrandale/clarksdale will be out for Christmas season

Quit being quite so pessimistic. Theres nothing new for you to compare, it'll be the same case and cooling design as we've had with bloomfield.

and you completely missed the fact im talking about i5.

And you completely miss the fact this presentation is glorifying a 32nm product for a 45nm product we have yet to see, that says its going to come out this year when the i5's are delayed? (and dude there are working samples, so intel is pushing them behind)

So intels great plan is to release 45nm i5's for less then 6 months, and then shoot themself with a 32nm transition?


Expecially with today's market recession?

 

[ilkhan]

I knew you were talking about i5 (aka lynnfield). It'll be out when its out. Hell, people aren't buying new chips right now anyway, why would intel want to spend the extra on making expensive nehalems vs inexpensive core2 chips anyway? 2 tiny chips are cheaper than 1 bigger chip for them. Same reason for the upcoming i7 950, they want to try and push people towards the higher profit chip, too many people are buying 920s right now compared to the 940.
And the plan is to use the mature 45nm process on the big chips (the quads) while perfecting the 32nm process on the duals (and the high dollar per chip gulftown which can absorb crappy yields while making a profit), before using the then mature 32nm process on the big chips again.

 

[Idontcare]

Originally posted by: aigomorla
So intels great plan is to release 45nm i5's for less then 6 months, and then shoot themself with a 32nm transition?

Expecially with today's market recession?

I'm not really following your logic on how/why this is a concern. To launch i5 now would require more 45nm capacity than Intel desires to purchase.

The planned transition to 32nm and investments there make more sense than to increase 45nm capacity now and then not need it in 6 months.

This would be a different story if Phenom II gave Yorkfield a problem. It didn't.

So i5 Lynnfield will release when it makes sense to do so, as 32nm capacity ramps and 45nm capacity gains some slack. The transition to 32nm is intended to be orderly, efficient, business intelligent.

There is a difference between delaying for purpose of fiscally driven market-timing reasons versus delaying because of technical issues limiting the physical delivery of the chips.

Sure Intel could ship i5 today, they could also give away 965 chips for $20 a pop if they felt so compelled, giving away 965's and releasing i5 are all actions not taken for the same legitimate business reason - maximizing shareholder value.

As I see it, they are judiciously waiting to launch i5 until they can time the shift in production from 45nm penryn to 32nm clarkdale, converting the 45nm capacity allocated to penryn into capacity allocated for lynnfield.

Optimizing their capex. Who can blame them for doing their jobs?

 

[aigomorla]

ok lets see.

32nm due date is november. Correct?

Lets see.

 

[dmens]

i5 platform total cost for intel should be less than c2q platform, so they should be cranked out as soon as they are approved.

it should be obvious to everyone what price points they'll be going after.

 

[Nemesis 1]

Originally posted by: dmens
i5 platform total cost for intel should be less than c2q platform, so they should be cranked out as soon as they are approved.

it should be obvious to everyone what price points they'll be going after.

Oh ya its as plan as day what Intel is up to here.

 

[Idontcare]

Originally posted by: aigomorla
ok lets see.

32nm due date is november. Correct?

Lets see.

Aigo how about a friendly wager

If Intel debuts 32nm before Dec 1, 2009 then you give me your i7 965, if 32nm does not debut by Dec 1, 2009 then I will give you your choice of either my B3 stepping QX6700 or slowspyder's 9850 phenom

(note: slowspyder, I may need to borrow your phenom come Dec 1st, don't worry I won't need it for long, just trust me )

That seems like a reasonable gentlemen's bet, right?

Now I'm a little slow (I mean mentally, can run like the dickens) so don't be surprised if it seems like you are taking advantage of me here by accepting this once in a lifetime offer. My wife tells me all the time not to make bets because I so rarely know the value of what I'm giving away :laugh:

 

[aigomorla]

Originally posted by: Idontcare

Now I'm a little slow (I mean mentally, can run like the dickens) so don't be surprised if it seems like you are taking advantage of me here by accepting this once in a lifetime offer. My wife tells me all the time not to make bets because I so rarely know the value of what I'm giving away :laugh:

LMAO...

the 965 is going bye bye next month.

on the 10th i get a 975 D0.

Ummmm... i'll wager you a B3 kentsfield.

i have a spare that i was trying to kill which i failed horribly at. Meaning, b3 kents are tanks.

 

[imported_Scoop]

I'm disappointed for no quadcore Westmere

I guess they want to sell those Lynnfields... oh well, I'll be waiting for Sandy Bridge.

 

[daw123]

Originally posted by: aigomorla

Originally posted by: Idontcare

Now I'm a little slow (I mean mentally, can run like the dickens) so don't be surprised if it seems like you are taking advantage of me here by accepting this once in a lifetime offer. My wife tells me all the time not to make bets because I so rarely know the value of what I'm giving away :laugh:

LMAO...

the 965 is going bye bye next month.

on the 10th i get a 975 D0.

Ummmm... i'll wager you a B3 kentsfield.

i have a spare that i was trying to kill which i failed horribly at. Meaning, b3 kents are tanks.

Aigo, how does it works with your (presumed) sponsor / supplier.

Do they send the latest chip to you and you have to return the previous one? Or do you get to keep the chips they give you? Do you have to pay for the ES CPUs?

And how do you get into this; is it by knowing people in the industry and by making a name for yourself as someone who knows a lot about the various facets of a PC?

I'm just curious. I don't know enough about PCs nor know anyone in the industry; hence it?s very unlikely this would ever happen to me.

Edit: I should start proof-reading my posts, so that I don't have to go back and correct spelling and punctuation. Plus, I always seem to add things to my posts to clarify my crappy explanations / arguments, so I don't get flamed.

 

[21stHermit]

Originally posted by: Idontcare
I'm sure the desktop and mobile 2c/4t chips are the same die with the mobile chips being put thru extra validation for the more aggressive power-saving functions versus desktop chips just having those functions fused off or microcoded to be ignored.

Could you expand on this thought.

Also, if they are the "same chip" how is the mobile able to use S989 whereas Clarksdale use S1156?

Thanks

 

[Rick James]

I'll take one of those 6 core, 12HT chips please