snb-e pushed back to January 2012

[ViN86]

That's a lot of juice.

[AngelGraves13]

If there's no USB 3.0 or Thunderbolt, I'll pass on the chipset and the CPU.

[shadow001]

Originally Posted by ViN86

That's a lot of juice.

To be honest, the chips do have an extra memory channel over the gulftowns(4 versus 3), and the AVX instruction set on all cores with sandy bridge E, while the gulftowns don't have it at all, and running AVX code uses more power and makes the chip run warmer, as it's already observed with the existing sandy bridge on socket 1155, but the E version brings 2 more cores to the party and still uses the 32 nm process as well as more cache onboard, so all these little things might add up in power use terms..


Ultimately though, if Sandy Bridge E is released in Q1 2012, might as well wait for the 22nm ivy bidge version wich is still stated as a Q2 2012 release, wich should use less power and perform better still clock for clock.....It should still use the X79 chipset boards as a basis for either one.

[ViN86]

Originally Posted by AngelGraves13

If there's no USB 3.0 or Thunderbolt, I'll pass on the chipset and the CPU.

Weird...

These two technologies will become standards. I would hope they would make an effort to add them. Looks like we will have to wait until Ivy Bridge or later.

[Bman212121]

Originally Posted by shadow001

Ouch if true:


http://techreport.com/discussions.x/21469


180 watts under load for the high end 6 core sandy bridge E chips, with intel recomending that PSU makers make their PSU's deliver at least 23 amps on the secondary 12volt rail dedicated to the CPU...I wonder how high it can get once overclocked....Yikes.


Ivy bridge with it's 22nm finfet process and 30% power savings over the 32nm process sounds rather nice right about now, and some here actually wanted an 8 core version Sandy bridge E while still on the 32nm process....

I really don't believe that at all... Intel already has working I7 980x cpus on 32nm... A die shrink isn't going to make it draw more power than the previous generation. It would be more likely that Intel can either bump up the clock speeds can keep the same TDP or use similar clock speeds with a lower TDP rating.

The one thing they might be mistaken on was some of the server cpus were supposed to have configurable TDPs. So rather than set the clock speeds based on a model number, each part would change based upon the BIOS setting. Maybe then I could see some BIOS settings letting them scale up to 180W draw to maximize speed.

This is an Anand article about Ivy bridge: http://www.anandtech.com/show/4378/i...nfigurable-tdp

If they did implement that I suppose SNB-E could hit 180W for a very short period, but that would still mean they would need some pretty high OC's to be able to do that. I'd have to imagine 3.6ghz hex cores would be possible in the same TDP, so it would be well into the 4ghz range with turbo modes.

[Viral]

Originally Posted by Bman212121

I really don't believe that at all... Intel already has working I7 980x cpus on 32nm... A die shrink isn't going to make it draw more power than the previous generation. It would be more likely that Intel can either bump up the clock speeds can keep the same TDP or use similar clock speeds with a lower TDP rating.

Not disagreeing, but SB-E is still 32nm, there is no die shrink.

For the record though, a die shrink can certainly make a chip draw more power than a previous generation if leakage is excessive. This is pretty much almost always a given for the first few steppings on a new node... usually they will continue with revisions until they get the leakage and other problems under control before releasing retail chips. All of that is irrelevant to this thread though

[Bman212121]

Originally Posted by Viral

Not disagreeing, but SB-E is still 32nm, there is no die shrink.

For the record though, a die shrink can certainly make a chip draw more power than a previous generation if leakage is excessive. This is pretty much almost always a given for the first few steppings on a new node... usually they will continue with revisions until they get the leakage and other problems under control before releasing retail chips. All of that is irrelevant to this thread though

My bad. Thought SNB-E was going to be on 22nm.

[Q]

Intel is just sandbagging again because there is no competition.

/wild speculation

[shadow001]

Originally Posted by Q

Intel is just sandbagging again because there is no competition.

/wild speculation

Pretty much this right here, and it turns out that even after all this wait, AMD's high end bulldozer even though it's better than previous AMD CPU's, still isn't enough to threaten the current high end intel chips, so there's no rush at all in releasing new high end parts....



Should be an interesting fight in the mid range market with current 4 core sandy bridge chips though...

[shadow001]

Originally Posted by slaWter

Ivy Bridge will be compatible to Sandy Bridge but not to Sandy Bridge E and X79.
There will be other, smaller 7-series chipsets for Ivy Bridge.





Intel confirmed a long time ago that TB and USB 3.0 will come with Ivy Bridge and the rest of the 7 series chipsets.

This is just another negative aspect of Intel's new release schedule.

There won't be a 22nm 6/8 core ivy bridge using the 2011 pin socket form factor for high end desktops?....I doubt Intel will exclusively limit it to socket 1155 and 4 cores for the mid range of the market...

[hirantha]

Sandy Bridge-E Coming November to a Store Near You Woot!!

[shadow001]

Originally Posted by hirantha

Sandy Bridge-E Coming November to a Store Near You Woot!!

If i'm reading this right, the chip to get would be the midrange 6 core model with the unlocked multiplier, even though it will be nearly 600$ at retail, and we still have to see the prices on the motherboards themselves too....