http://www.hardocp.com/article/2013/08/30/intel_ivy_bridgee_core_i74960x_ipc_ocing_review/

 

 

 

The general feeling I come away from the Ivy Bridge-E is that it is a good processor, but nothing stellar when compared to the Sandy Bridge-E predecessor. Is it better? Yes. It is a whole lot better? No. Should you ditch your Sandy Bridge-E system for a shiny new Ivy Bridge-E? No. If you happen to be coming from a sub-4GHz LGA 1366 "Gulftown" system, the Ivy Bridge-E is going to be a big step up. However if you are lucky enough to still be sporting a ~5GHz Gulftown processor, you are likely better off sticking with it.

 

As for overclocking, they got it to 4.6 GHz. In spite of it using solder instead of TIM like Ivy Bridge and Haswell, they were unable to stabilize the system at 4.7 GHz.

 

For most users, it looks like a 4770K @ 4.4 - 4.6 GHz would provide better performance. The main advantage with IB-E would be more PCI-E lanes (triple-SLI) and up to 6 cores/12 threads with the more expensive CPUs.

i thinking about to upgrade to a ivy-e to have 2 cores more for terrain loading.

 

because i still have blurries.

 

but i´m not sure if it`s worth it ?

 

 

 

 

Gesendet von meinem iPad mit Tapatalk HD

http://www.hardocp.com/article/2013/08/30/intel_ivy_bridgee_core_i74960x_ipc_ocing_review/

 

 

 

 

As for overclocking, they got it to 4.6 GHz. In spite of it using solder instead of TIM like Ivy Bridge and Haswell, they were unable to stabilize the system at 4.7 GHz.

 

For most users, it looks like a 4770K @ 4.4 - 4.6 GHz would provide better performance. The main advantage with IB-E would be more PCI-E lanes (triple-SLI) and up to 6 cores/12 threads with the more expensive CPUs.

 

 

Solder under the IHS may actually be a BAD thing.

 

May be that delidding a chip with TIM under the IHS offers lower temps that a soldered chip.

Solder under the IHS may actually be a BAD thing.

 

May be that delidding a chip with TIM under the IHS offers lower temps that a soldered chip.

 

I don't know - They were seeing 65C full load at 4.5 GHz and 1.31V. That's hard to beat even if you do a perfect job with the delid.

 

 

Solder under the IHS may actually be a BAD thing.
 
May be that delidding a chip with TIM under the IHS offers lower temps that a soldered chip.

 

How so Martin?

How so Martin?

 

I'm thinking that perhaps, given how thin the liquid pro is spread between the IHS and die when we de-lid, that it may offer somewhat better thermal transfer than the solder applied to Ivy E.

 

I know solder has excellent thermal transfer, but how thick the layer is impacts heat conduction also. Solder will be quite thick, liquid pro super thin.

 

Same applies to chips like Ivy and Haswell with the Intel TIM. It's the thickness of the layer that's the real issue, not really the quality of the TIM.

 

Just thinking aloud really, probably nonsense.

 

I was reading recently in a de-lidding forum, that someone has de-lidded a soldered chip, think it was a 2500K IIRC. The solder was quite soft, and he didn't damage the die. Not sure about the temp difference.

Yes, Liquid Pro may in fact be even better than solder. When you delid, you're also removing the rubber seal that the IHS rests against. This moves it very close to the CPU die. It's so close that a "painted on" layer of Liquid Pro is enough.

 

However in the real world, I think the difference between solder and CLP is quite insignificant compared to the difference between TIM and solder.

 

 

It's the thickness of the layer that's the real issue, not really the quality of the TIM.

 

I don't know Martin.  Regardless of thickness in the end it's all about just how much heat does transfer thru the medium over what period of time.  Other factors such as degree of contact for example would come into play as well as layer thickness.   A thicker layer may actually, due to better total surface adherence and excellent conductance of the solder media, suck off more heat than Liquid Pro over the same period of time as the solder itself is a heat sink.

Agree.

 

Who knows.

 

Unless some brave fool has a go at de-lidding Ivy E and manages to do so successfully [almost impossible I would have thought] we will never know.

 

A Chinese guy did it, and killed the chip.

 

From what I've heard, Intel are using a strong epoxy solder to fuse die to IHS.

 

I believe epoxy solder cures at room temperature. So in essence, they are using GLUE.

 

I don't think a glue will be as efficient as conventional solder, despite the fact it's impregnated with silver.

JimmiG, on 01 Sept 2013 - 06:57 AM, said:

I don't know - They were seeing 65C full load at 4.5 GHz and 1.31V. That's hard to beat even if you do a perfect job with the delid.

65C is about what I see w/ SB-E at full load using Prime 95 w/o HT so that seems quite in the ball park if you are saying this about IB-E.

JimmiG, on 30 Aug 2013 - 12:41 PM, said:

The general feeling I come away from the Ivy Bridge-E is that it is a good processor, but nothing stellar when compared to the Sandy Bridge-E predecessor. Is it better? Yes. It is a whole lot better? No.

It *should* be ~5% better if it parallels quad core versions AND it should support PCIe 3.0 that SB-E really doesn't support. When I assembled my current machine I needed to try to decide if <= 5% improvement plus PCIe 3.0 was worth waiting for and decided for all practical purposes wasn't worth the wait. If I kill this lovely SB-E chip by overvolting I'm hoping my X79 motherboard will accommodate IB-E which it is purported to do I assume thru a BIOS update. Great news on IB-E and solder ;o)

It could be great news on IB-E and solder. Unless what I've heard is true, and that they have used an epoxy solder.

 

In which case it's pretty much glue. I don't know much about epoxy solder, but I have doubts it's anywhere near as good as conventional solder.

 

The problem is that there's less metal in the epoxy solder than conventional solder, so the weaker the electrical flow. Not sure how it compares regarding heat conduction but I suspect not as good.

 

The advantage of course is no heat required.

 

Something tells me it's probably not a solder at all, more like a thermally conductive epoxy.

Something tells me it's probably not a solder at all, more like a thermally conductive epoxy.

 

Yes, but from what Jimmi says, which IMO says it all really:

 

JimmiG, on 01 Sept 2013 - 06:57 AM, said:

I don't know - They were seeing 65C full load at 4.5 GHz and 1.31V.

 

With a TDP of 130W, it's really doing pretty much exactly what it should do w/ solder if SB-E is a reasonable comparator for solder as an interface material near as I can see.  When I run Prime95, not FSX, w/o HT enabled, I'm seeing right around 68F (w/o my A/C unit on but using the DH-14), but that is at 4.4Ghz and 1.305v set in the BIOS, VID around 1.275.   So really, 65C at full load a 1.31v seems right in there.  Were this a vastly inferior interface material I don't think you'd be seeing anywhere near this, unless I'm misinterpreting this perhaps, i.e., if they're referring to case temp not core temp for example.  I'm talking core temp for my SB-E report.

 

Not being able to get stability over 4.7Ghz is pretty hard to comment on as there is so much to consider--for example, how many samples did they try with, etc.  It seems there is generally a little less overclockability w/ all tri-gate processors, emphasis on 'little'!

65c was with a full koolance water cooling loop I believe, would be quite a bit warmer with air cooling, Your D14 for example. Perhaps 75-80c

 

At 4.5 even ivy or Haswell with TIM could cope. It's above a particular voltage threshold that TIM (or perhaps less efficient epoxy solder) becomes less of a thermal conductor and more of an insulator.

 

TIM or solder or epoxy is only really relevant above the critical voltage threshold. For my Ivy that threshold is 1.35 volts.

 

I'm playing devil's advocate to a degree hear. But perhaps, 4.7 would have been achievable for Ivy E with old fashioned solder or even Aa thin layer of liquid pro TIM

This is the cooling they used...

 

The cooling system we are using is the Koolance EX2-755 (Exos-2 V2) with a Koolance CPU-380i water block that works with all recent generation socket configurations. To adapt this 775/1155/1156/1150 block to your LGA 2011 or LGA 1366 system you will need these adapter posts.

 

 

The Bottom Line

 

 

 

 

The general feeling I come away from the Ivy Bridge-E is that it is a good processor, but nothing stellar when compared to the Sandy Bridge-E predecessor. Is it better? Yes. It is a whole lot better? No. Should you ditch your Sandy Bridge-E system for a shiny new Ivy Bridge-E? No.

65c was with a full koolance water cooling loop I believe, would be quite a bit warmer with air cooling, Your D14 for example. Perhaps 75-80c

 

At 4.5 even ivy or Haswell with TIM could cope. It's above a particular voltage threshold that TIM (or perhaps less efficient epoxy solder) becomes less of a thermal conductor and more of an insulator.

 

TIM or solder or epoxy is only really relevant above the critical voltage threshold. For my Ivy that threshold is 1.35 volts.

 

I'm playing devil's advocate to a degree hear. But perhaps, 4.7 would have been achievable for Ivy E with old fashioned solder or even Aa thin layer of liquid pro TIM

Wow that's terrible!  Might as well have used styrofoam to hold down the IHS!  Geez that makes me want to pick up another 3930K before they are no longer available!  Martin, see if you can find a comparison of cooling solutions with the Koolance EX2-755 in it.  Couldn't find one!