Nehalem prices leaked out?

[MarcVenice]

http://www.hkepc.com/?id=1447&fs=c1h

No clue how to translate it, but it shows a nice graphic of the 2.66ghz bloomfield CPU for $283. I wonder what a s1160 mobo will cost?

Hope this isn't a repost ...

 

[Hugh H]

Yep, it was reported a few days back... keep in mind that that is the lot price (like, for a 1,000 of them), actual retal price would be around $312.00 for the 2.66Ghz version.

I am more concerned about the total cost when you include the new motherboard, RAM, heatsink, etc...

Also, it would be nice to know if these new sockets will support the 32nm Nehalems when they come out... since these 45nm versions, although definitely packing more power, run very hot when compared to 45nm Penryns. I could buy a 45nm Nehalem when they first come out, and then upgrade to a 32nm one if my motherboards supports it.

Otherwise, I would skip the 45nm Nehalems completely and wait for the 32nm version.

 

[BLHealthy4life]

Originally posted by: Hugh H
Yep, it was reported a few days back... keep in mind that that is the lot price (like, for a 1,000 of them), actual retal price would be around $312.00 for the 2.66Ghz version.

I am more concerned about the total cost when you include the new motherboard, RAM, heatsink, etc...

Also, it would be nice to know if these new sockets will support the 32nm Nehalems when they come out... since these 45nm versions, although definitely packing more power, run very hot when compared to 45nm Penryns. I could buy a 45nm Nehalem when they first come out, and then upgrade to a 32nm one if my motherboards supports it.

Otherwise, I would skip the 45nm Nehalems completely and wait for the 32nm version.

if the 32nm Hehalems are the same socket, it is very very likely that the mobo will support them.....usually only a BIOS upgrade is needed...

 

[videogames101]

http://translate.google.com/tr...1GGLQ_ENUS285%26sa%3DG


Translated.

(Google FTW)

 

[harpoon84]

Originally posted by: Hugh H
Yep, it was reported a few days back... keep in mind that that is the lot price (like, for a 1,000 of them), actual retal price would be around $312.00 for the 2.66Ghz version.

I am more concerned about the total cost when you include the new motherboard, RAM, heatsink, etc...

Also, it would be nice to know if these new sockets will support the 32nm Nehalems when they come out... since these 45nm versions, although definitely packing more power, run very hot when compared to 45nm Penryns. I could buy a 45nm Nehalem when they first come out, and then upgrade to a 32nm one if my motherboards supports it.

Otherwise, I would skip the 45nm Nehalems completely and wait for the 32nm version.

I'd agree over the platform cost, the CPU price is reasonable but the mobos will be fairly expensive and DDR3 ain't cheap atm either.

I'm not sure what you mean by Nehalems running very hot though, it consumes ~10% more power clock for clock than Penryn but its also a lot quicker clock for clock, I'd say it even holds a performance/watt advantage over Penryn.

http://www.anandtech.com/cpuch...howdoc.aspx?i=3326&p=8

 

[PeteRoy]

I'm guessing It takes more power because it has integrated memory controller no? So it means the chipset consumes less.

 

[harpoon84]

Originally posted by: PeteRoy
I'm guessing It takes more power because it has integrated memory controller no? So it means the chipset consumes less.

I'd imagine so, I think HT has something to do with it too, as the cores are kept 'busier' when under full load.

 

[Idontcare]

Originally posted by: PeteRoy
I'm guessing It takes more power because it has integrated memory controller no? So it means the chipset consumes less.

There's a number of entirely improper comparisons to make between Yorkfield and Nehalem that will all result in the correct answer

For example, we could argue that because there are more logic xtors and less cache xtors on nehalem versus yorkfield we should expect nehalem to consume more power at the same clockspeed. But the list of reasons why such an analysis is flawed would probably exceed 20 or 30 distinct reasons.

However such simple rules of thumb tend to work despite the very good reasons they shouldn't (for one total xtors means nothing about regarding voltage and amps used at any given instance during computing)

So I would argue that yes Nehalem will consume more power at same clockspeed than a Yorkfield...it has an IMC and less cache/more logic xtors.

 

[stuff311]

Originally posted by: Idontcare
So I would argue that yes Nehalem will consume more power at same clockspeed than a Yorkfield...it has an IMC and less cache/more logic xtors.

And you would be correct.

 

[bunnyfubbles]

Which is why Intel liked to pimp out the new marketing term "performance per watt" starting with the release of C2D...as long as performance outpaces power consumption and heat production, we're golden. However I'm sure there will be plenty of disappointed enthusiasts if these chips can't hit magical numbers like 4GHz, even if they're getting 4.5GHz Penryn performance out of a 3.8GHz Nehalem.

 

[Tempered81]

Originally posted by: videogames101
http://translate.google.com/tr...1GGLQ_ENUS285%26sa%3DG


Translated.

(Google FTW)

"Socket 1160 shipments will surpass the old Socket 775, officially entered Shidaijiaoti."

ahhh, okay.

Seriously, though: what exactly is preventing us from buying the low-end models and overclocking them like we're doing with penryns? Is it that the bus speed is always 133mhz, and the multiplier is unchangeable, except for the extreme models?

 

[Acanthus]

Originally posted by: BLHealthy4life

Originally posted by: Hugh H
Yep, it was reported a few days back... keep in mind that that is the lot price (like, for a 1,000 of them), actual retal price would be around $312.00 for the 2.66Ghz version.

I am more concerned about the total cost when you include the new motherboard, RAM, heatsink, etc...

Also, it would be nice to know if these new sockets will support the 32nm Nehalems when they come out... since these 45nm versions, although definitely packing more power, run very hot when compared to 45nm Penryns. I could buy a 45nm Nehalem when they first come out, and then upgrade to a 32nm one if my motherboards supports it.

Otherwise, I would skip the 45nm Nehalems completely and wait for the 32nm version.

if the 32nm Hehalems are the same socket, it is very very likely that the mobo will support them.....usually only a BIOS upgrade is needed...

That is not an absolute, the VRMs also have to support new power requirements should they change.

 

[Idontcare]

Originally posted by: jaredpace
Seriously, though: what exactly is preventing us from buying the low-end models and overclocking them like we're doing with penryns? Is it that the bus speed is always 133mhz, and the multiplier is unchangeable, except for the extreme models?

From my meager understanding there isn't anything about the QPI bus that makes it any more robust against the analogous overclocking techniques that currently work for overclocking AMD systems.

 

[taltamir]

Originally posted by: bunnyfubbles
Which is why Intel liked to pimp out the new marketing term "performance per watt" starting with the release of C2D...as long as performance outpaces power consumption and heat production, we're golden. However I'm sure there will be plenty of disappointed enthusiasts if these chips can't hit magical numbers like 4GHz, even if they're getting 4.5GHz Penryn performance out of a 3.8GHz Nehalem.

Intel moved the FSB to the chip itself and used a secondary clock to set the speeds... there is no known way to OC these.
Maybe some mobo maker will find a way to do so later on, but as of right now they appear to be non OCable.

The 2.66 @ 284$/each per 1000 units is the only one that isn't extremely priced. After some margins they would probably go for 350$ a piece. Since nehalem is about 20-30% faster per clock...
2.66 x 1.3 = 3.458
So a 4ghz penryn would outperform it.

the 3.2 ghz model though... 3.2 * 1.3 = 4.16
And since it is an extreme version it will probably have an unlocked multiplier, making it ocable.

It will shatter the high end, but it will not replace the low, medium, and medium-high when it is released.

Originally posted by: Idontcare

Originally posted by: jaredpace
Seriously, though: what exactly is preventing us from buying the low-end models and overclocking them like we're doing with penryns? Is it that the bus speed is always 133mhz, and the multiplier is unchangeable, except for the extreme models?

From my meager understanding there isn't anything about the QPI bus that makes it any more robust against the analogous overclocking techniques that currently work for overclocking AMD systems.

Clock speed is determined by a second internal clock AFAIK. So you might be able to OC the bus only (increasing bandwidth) with the same techniques as used for AMD, but it will not OC the actual CPU.

 

[Idontcare]

Originally posted by: taltamir
Intel moved the FSB to the chip itself and used a secondary clock to set the speeds... there is no known way to OC these.

Clock speed is determined by a second internal clock AFAIK.

I find it very difficult to believe that Intel would go to such costs just to prevent a very minor percentage of total users from overclocking what is otherwise an already low-margin chip.

Possible? Yes. Plausible? No.

Were 50% of $100 chips overclocked, undermining the sales of higher margin chips then I could see the justification for increasing the cost structure of all those $100 chips by integrating a second clock gen (and all the accompanying logic).

But what's the cost structure justification when perhaps at most 2% of those $100 chips are overclocked at the expense of the consumer not purchasing a higher clocked higher margin CPU?

 

[taltamir]

well.... the only difference between their 250 and 1500$ is that the 1500 is clocked higher by default and has an unlocked multi... an extra 1250$ per sale in their pocket. how much do they make on the 250$ chip? 125$ is my guess. making 11 times the margin on one chip sale. Maybe they think it will be worth it to make all chips a little more expensive to make... How much would it even cost to add such a secondary internal clock? if only 5$, then a single sale of 1500$ chip would pay for such limiters in 250 CPUs. 1$? then a single such sale pays for 1250 CPUs.

And maybe they are worried about warranty... besides, this is essentially DRM for your processor, and DRM never had anything to do with logic and making money. (aka, it costs you money and doesn't net you any to use it, a purely emotional knee jerk response)

And it is all rumors right now, we will know about ocability when the chips are out.

 

[aigomorla]

lol i love hypothesis thread.

 

[boomhower]

The chips ain't bad but as others have eluded to the whole system costs is going to be pretty bad. The X58 boards are going to be more than the current X48, there is $300 compared to the $150 P45's we enjoys today. Since we have gotten so used to running 4GB's in Vista now you are going to have to drop to 3GB or set up to 6GB's for tri-channel, again another $300+. Then add a new HS and your $300 chip got very expensive, very quickly.

 

[taltamir]

300$ chip? I don't know about you, but if I was in the business of selling stuff I wouldn't buy a rapidly depreciating product for 284$ a unit in sets of 1000, and sell them for 300$ a pop...
For perspective, that is purchasing at 284000$ and selling at 300000$ total... and that assumes no price cuts are made, and doesn't take into account other forms of loss...
I forsee it at 350$ a piece minimum...

 

[boomhower]

Then it is a good thing you are not in the CPU business. Unless something has changed recently the official tray price for the E8400 is $183. It is going for $179.99 on newegg. It has been like this for quite some time. At first they will price gouge them and they will settle right at the tray price.

 

[harpoon84]

Originally posted by: taltamir
300$ chip? I don't know about you, but if I was in the business of selling stuff I wouldn't buy a rapidly depreciating product for 284$ a unit in sets of 1000, and sell them for 300$ a pop...
For perspective, that is purchasing at 284000$ and selling at 300000$ total... and that assumes no price cuts are made, and doesn't take into account other forms of loss...
I forsee it at 350$ a piece minimum...

Big E-tailers like Newegg probably buy well in excess of 1000 units and will undoubtably get a discount in the process.

Look at their prices compared to the 1K price, theres not much difference, maybe +5%.

Then again, I wouldn't be surprised to see price gouging at the beginning, like the E8400s that were going for $250 at one stage.

 

[taltamir]

newegg sales a lot, but i very much doubt they can quickly move thosands of 300$+ CPUs that require 300$+ motherboards and 300$+ ram...
This isn't a long awaited 200$ video card after being stuck with the same high end card for 2 years kind of case.

Ofcourse, if DDR3 prices go down by then, and boards are availble at 150 instead of 300. then I predict them selling like hotcakes.

Originally posted by: boomhower
Then it is a good thing you are not in the CPU business. Unless something has changed recently the official tray price for the E8400 is $183. It is going for $179.99 on newegg. It has been like this for quite some time. At first they will price gouge them and they will settle right at the tray price.

Unless newegg is selling them for a loss of 3 dollars each, then the tray price is either a lie, or they get a much bigger discount at higher volumes.
This is still very good to hear though, I would love for them to settle down to 280$... well, I would love that if they are overclockable, otherwise I would just get a q9300 and OC it to 3.5+ghz and end up with a faster processor.