Ivy Bridge Temperatures Could Be Linked To TIM Inside Integrated Heatspreader: Report

jaredpace said:
If you have a good CPU watercooling setup you should definitely get the 3770K versus Sandy because the only limitation is Temperature.
Click to expand...

Or high end cooling/all in one water setups.
Even with my SB OC, I wish I would of went with either of the high end units for better temps.....mine is at around 67 with all cores at max.
 
As an engineer who took thermodynamics and now working with heat exchangers, this is my opinion on how Intel could fix the problem.

When you want to increase efficiency you increase the heat disapation area. With a standard heat exchanger, you either increase the number of fins per inch, or increase the thickness of a fin, and increase it's length. The former is more effective. But the later works as well to an extent.

Since we can't do the first (increase the number of fins, as there is only one fin (the heat spreader), we have to increase it's ability to transfer heat over longer distances (through thickness)

First off, you don't increase the total thickness of the just the top layer above the chip. That could work against you as you increase distance from heat source to disappation/exchange area.

So you increase the thickness to the sides of the chip. You do this by creating a IHS with a slot carved out, very close to the size of the IB die. Then stick fluxless solder in that slot, mount it to the surface, and then melt it to the core. This will aide in dispearing heat more evenly across the entire surface area of the chip.

fluxless.jpg
 
Does anyone have a reference for Intel officially admitting high power density as an issue here? It would make useful for a proposal I am writing up right now (related to electronic cooling).
 
blade52x said:
Does anyone have a reference for Intel officially admitting high power density as an issue here? It would make useful for a proposal I am writing up right now (related to electronic cooling).
Click to expand...

There's a quote floating around - it was in an article at Tech Report initially, but I've seen it other places now as well.
 
Sure Intel could have squeezed more cooling efficiency by using a more efficient TIM from what I have gathered so far. In my opinion they should have even if it cost a little more. Quality goes a long way in selling a product and you therefore achieve a better reputation and a reasonable excuse to ask for more selling your product. I am all about that concept. Unfortunately today's world is filled with good ideas but poorly implemented because of greed or just plain lack of caring.

What I find interesting about the new Intel chip is people can't seem to wrap their head around the fact that Ivy is actually faster at lower frequencies using less power than Sandy. People assume in order to be the winner you need to have a higher frequency. While higher frequency does improve performance over time Ivy performs equally or better and more efficiently at say 4.6 GHz versus Sandy at 4.8 GHz. If you take Ivy to 4.8 GHz then Sandy needs 5 GHz in order to keep up and yet it is still less efficient doing the same amount of work. Ivy does create more heat because it is doing the same amount of work in less space and the lack of work area inhibits cooling off the process as efficiently. Ivy has a higher thermal limit so a higher limit to survive the heat. You can attempt to help control Ivy's heat by having an enriched environment (a strong chip, configurable high performance Mobo, efficient cooling). You can also change your approach to overclocking by reducing electrical input (thereby reduce heat energy) for a given amount of frequency you would normally need to use on Sandy. If you can dissipate the heat fast enough, for example an effective cooling solution, then you can enjoy even more from Ivy. Ivy is by no means a large leap from Sandy but because of Ivy's reduced size and processing facility it's enough to claim the top position in it's class. As I'm sure future versions of this platform will also be an improvement I just hope Intel (or anyone for that matter) puts the time and effort into a higher quality product so that it can be the best that it could possibly be.
 
it is my understanding, from a paper I read that polymer based TIM paste breaks down over time and gradually looses its conductivity. That is OK for a CPU cooler where you can change the paste, but maybe not under the IHS where you can't change it. Perhaps Intel choose a paste that had a longer longevity at the acrifice of some conductivity.
 
pwrpc said:
What I find interesting about the new Intel chip is people can't seem to wrap their head around the fact that Ivy is actually faster at lower frequencies using less power than Sandy.
Click to expand...

A sensible observation, hence

1. For non-overclocker, go Ivy Bridge, you get the stated benefit, your maximum Processor temp will be within spec since you cannot raise volt and clock to reach 4.5GHz. When you stay within the spec, most of the time it will be sensible and reasonable. Proof --> many server processors run for years without anybody worrying about performing thermal paste cleanup, just need to clean dust,fans and other necessary within the case, if you insist. Check with your data centre server admins with racks and racks of servers, I believe most of the time they never need to pull the servers out to perform this procedure on yearly basis.

2. For determined overclockers, if you have reservation per the temp, then go Sandy Bridge. Most overclockers will pair their rigs with discreet GPU so Intel HD improvement is of little concern to them. You can perform manual processor thermal paste cleanup and replacement easily on periodical basis with minimal concern, if you insist.

2.1 Unless you want to overclock and also activate Lucid Virtu MVP, then it is a more fluid situation since HD4000 does bring DX11 and maybe Lucid software improvement to up the FPS. However, I see less active attention given to Lucid in this overclocking forum.

Cheers
 
GeneO said:
it is my understanding, from a paper I read that polymer based TIM paste breaks down over time and gradually looses its conductivity. That is OK for a CPU cooler where you can change the paste, but maybe not under the IHS where you can't change it. Perhaps Intel choose a paste that had a longer longevity at the acrifice of some conductivity.
Click to expand...

Or perhaps they Chose a paste that breaks down even faster over time. XD
 
You must log in or register to reply here.
Back
Top