New Core I7 Cpu's In the Future??

[RaptureMe]

Just wondering if there is going to be more Core I7 cpu's to choose from in the near future besides whats out now??
I am not interested in Core I5 just Socket 1366 CPU's only..
Also wanting to know is this the new Intel Server socket Replacing s775 and s771??
So will we see Xeons on the 1366 socket then??
Thanks for the help..

 

[ajaidevsingh]

Well a shrink of the i7 is expected. Since i7 runs quite hot as it is one may benefit a lot with a shrink.

I think the server based on i7 architecture will run on Socket 1366. The main difference of the two is an added QPI path along with other additions.

 

[dmens]

it is amusing when people say i7 is "hot". quad penryn's were just as hot. i7 just has sensors in better places.

 

[ajaidevsingh]

Conroe were hot as compared to Penryn and because of the shrink Penryn can run faster and cooler.

Also i7 does run hotter than a penryn quad.

 

[RaptureMe]

Well really I just wanted to know as I just jumped ship from the system in my sig to an I7 rig.
I was hoping that since I went ahead and went the 1366 route that there was going to be a viable upgrade path.
It would suck balls if 1366 was a special limited edition socket and was only going to be open for the 3 cpu's out there right now and no others were going to be realeased.
For hardcore upgrade junkies like me that just would not stand!!
I am hoping for an octo core with a total of 16 threads for the end of 2009 be it server or not it would be going into my I7 Mobo..

 

[CrazigXn]

octo core with 16 threads would be a beast with 24 gigs of ram

 

[RaptureMe]

Yeah Octo Core with 16 threads and 24GB ram all just to surf the internet.
NO NO J/K LOL Its being used for Video Editing,Compression,Surfing,Burning,Gaming,File Server ect..

 

[Idontcare]

Originally posted by: ajaidevsingh
Also i7 does run hotter than a penryn quad.

Based on what data? Data generated by the internal DTS? If so then you are failing to take note of Dmens post. (hint - dmens works/worked on Nehalem development...he sometimes knows what he's talking about...)

Also DTS temps on Penryn (at least, maybe i7 too) are not calibrated to be accurate at any temperature other than TJmax.

So I am really curious as to what basis you are qualifying your claim.

 

[pm]

For what it's worth, I've worked on chips like that - where on one generation due to floorplanning or bad power estimation, the sensors are not in the optimal location (or even vaguely close to the best location) and then it's moved on the next generation and people think the new chip is hotter than the previous when in fact, the opposite was true. Not going to name any names... but I can think of at least one project where that happened.

You might think that moving the sensor a couple of millimeters wouldn't make a big difference, but if most of the power is being consumed in the integer execution unit and you put the sensor next to the L2, the readings will be a lot lower than you would think given the distance involved.

If you want to estimate power for a given product vs. another a better choice is a clamp ammeter on the CPU power rails.


* Not a spokesperson for Intel Corp.

 

[Tempered81]

so yorkfield & bloomfield "run equally hot" is the verdict?

 

[pm]

Regarding the OP's topic, I thought this entry at Wikipedia was interesting:
http://en.wikipedia.org/wiki/I...em_(microarchitecture)
scroll down to the table with the sockets.

I don't vouch for it's accuracy in any way (I actually don't know the answer, I work in an entirely different part of the company...).


* still not an Intel spokesperson...

 

[Tempered81]

Looks like you might get lucky if a 6core 32nm westmere works on your motherboard a year from now.

 

[dmens]

Originally posted by: jaredpace
so yorkfield & bloomfield "run equally hot" is the verdict?

i'd say that is a trick question since hot regions tend to be localized.

imho the answer is, does it really matter how hot the part is if the part is stable at your desired overclock?

i personally am far more concerned with the high voltage that is causing the heat to begin with. heat itself won't damage the part, even at 100C, afaik.

 

[RaptureMe]

PM thanks for the link thats what I was looking for.
I thought s1567 was the one getting 6 cores with 12 threads and teh s1366 was geting 8 cores with 16 threads??
I thought I read that here on A/T forums that 8/16 was headed for the main server socket which if I'm not mistaken is infact s1366??

 

[VirtualLarry]

Originally posted by: dmens
heat itself won't damage the part, even at 100C, afaik.

Thanks for stating that. It shows that I'm not crazy running my E2140s at up to 9C from TJmax. They are rock-solid stable.

 

[evilbix]

Originally posted by: RaptureMe
PM thanks for the link thats what I was looking for.
I thought s1567 was the one getting 6 cores with 12 threads and teh s1366 was geting 8 cores with 16 threads??
I thought I read that here on A/T forums that 8/16 was headed for the main server socket which if I'm not mistaken is infact s1366??

They have 6-core already, well.. according to newegg. I don't know much more about it, but appears to be socket 604

http://www.newegg.com/Product/...x?Item=N82E16819117176

*edit* The 6-core is penryn based, nvm

 

[IntelUser2000]

Just wondering if there is going to be more Core I7 cpu's to choose from in the near future besides whats out now?? I am not interested in Core I5 just Socket 1366 CPU's only.. Also wanting to know is this the new Intel Server socket Replacing s775 and s771?? So will we see Xeons on the 1366 socket then?? Thanks for the help..

Q: Will there be more i7's for higher end??
A: Yes, faster i7's are already on the roadmap. At least 3.46GHz is on the horizon, but depending on our luck we might see a 3.73GHz one. They are probably all EE versions though.

Q: Xeons on socket 1366, yes or no?
A: Yes it looks like they'll have same pin count, but I doubt the compatibility will be there. The Gainstowns have extra QPI link and support for SMP, while Bloomfield doesn't.