http://www.fudzilla.com/home/item/31064-haswell-e-supports-ddr4

 

 

Haswell-E has support for DDR4 & up to 12 or 16 cores.  

and some more about OC

 

http://www.anandtech.com/show/6898/intel-details-haswell-overclocked-at-idf-beijing

and some more about OC

 

http://www.anandtech.com/show/6898/intel-details-haswell-overclocked-at-idf-beijing

Unfortunately, nothing too meaningful: 

 

'What Intel isn't talking about publicly is just how overclockable Haswell is. I'm hearing great things about LN2 overclocking, but other than the high-end memory vendors being pleased with the new platform I haven't heard much on what will be possible on air.'

A few people are saying that Intel will return to using fluxless solder for Haswells IHS instead of TIM which should mean better overclocks vs Ivy Bridge but I don't know if these folks are well informed or just wishful thinkers.

[delete]

Get more work done per CPU cycle.....Haswell.

 

HLJAMES

With the recent news about desktop sells, I dont understand why theses people are still building CPU's optimized for them. I would be optimizing for portable devices which means power conserving way more than speed.

RWFeldman, on 15 Apr 2013 - 12:02 PM, said:

 

With the recent news about desktop sells, I dont understand why theses people are still building CPU's optimized for them. I would be optimizing for portable devices which means power conserving way more than speed.

I think they are definitely doing just that: SoC, low TDP, integrated GPU and voltage regulation, and so forth. Across the big world there is still enough demand for very high end parts so they are still trying to meet that demand too. I'm looking at this bad boy for my next build, where graphics is a smaller portion of the total processing power. I am building for XPlane 64bit or something else that can access what Haswell E will bring, or so I hope:

 

According to Fudzilla, the Haswell-E will have from 12 to 16 cores, a

TDP

of 130 W, and support for DDR4 memory that promises to deliver exponentially more bandwidth than previous generations of processors.

Right, just like Ivy Bridge E has support for "up to" 12 cores yet ships with 6 active  :rolleyes:

 

Fudzilla reports every rumor they get their hands on and provides no analysis to qualify the rumors nor discernment to say whether they should even report it in the first place. In other words: don't believe anything you read at Fudzilla unless someone else also reports on it and doesn't cite Fudzilla as the source.  

Right, just like Ivy Bridge E has support for "up to" 12 cores yet ships with 6 active

 

Where did that rumor come from, ie IB 12/6?  Thanks for the reply re Fudzilla I hadn't heard anything one way or the other about that source.  

 

Aren't some of the smartphones/tablets now coming w/ 8 core processors?  If so I think that's just testimony about the efficiency of multicore approaches.

Where did that rumor come from, ie IB 12/6?  Thanks for the reply re Fudzilla I hadn't heard anything one way or the other about that source.  

 

Aren't some of the smartphones/tablets now coming w/ 8 core processors?  If so I think that's just testimony about the efficiency of multicore approaches.

 

Ivy Bridge E 12 core rumors are all over the place, just Google it.  

 

There aren't any true 8 core phones yet, the "8 core" Samsung phones on the horizon have 4 primary cores that handle all the heavy lifting and 4 low power cores that run when the phone doesn't need to perform any intense calculations.  

How about this bad boy for XPlane or future multi-core exploiting simulators:  dual 8-core XEON processors:

 

http://www.tomshardware.com/reviews/xeon-e5-2687w-benchmark-review,3149.html

I built a system using two of those processors for a customer last year to replace his aging Mac Pro workstation.  It's fast, something like 10x faster for his rendering jobs.  Still not as fast as a heavily overclocked quad core for gaming or FSX though.  

 

 

Still not as fast as a heavily overclocked quad core for gaming or FSX though.  

 

I'm done w/ caring about designing a system to push dated code using crude approaches like aftermarket water cooling just to ramp up clock speed to a magical 5Ghz, when it gets you just a few more frames per second and ultimately will never really run FSX so that you can fly anywhere you want in any conditions in any 3rd party airplane and have it run fully.  This is possible w/ modern code approaches, i.e., so I'm building for next gen 64 bit simulation, be it XPlane or something else.  I'm sure the rig you built will outperform my 5 y/o Q9650 by a meaningful margin, but even if it didn't, I don't really care.   FSX runs adequately for me already, anything more is bonus.  I want to support smart design as there is much more upside in it over FSX.

It's not as simple as all that, and exotic cooling methods will always have their purpose. More performance is never a bad thing, after all. And for the record, I mow achieve a worst case scenario of 24 FPS on my PC with highest settings in FSX (except bloom and scenery shadows) and that's at 2560x1440 with 4x SGSS on top so yes, it is possible to have usable performance thanks to overclocking that is not otherwise achievable.

 

To the main point: Making software run faster via parallelization is a non-trivial problem. Not all tasks are able to be executed in parallel to other tasks due to data dependencies. Of those tasks which can be parallelized, not all of them stand to gain any performance from it. I'm as much for advancement as the next guy, but single thread performance is still critical and likely always will be.