Ivy Bridge thermal grease is the culprit for heat problems after all?
Ryan Martin / 4 years ago
Cast your mind back to late April and you will remember how the extra layer of thermal paste in the Ivy Bridge CPU construction was believed to be the culprit for high temperatures and this was followed by testing in China on the i7 3770k which then showed that the thermal paste made no difference. However, following even more testing initial thoughts are now proven to be true, that it is indeed the poor quality thermal paste put in by Intel that is causing problems. Extensive testing has continued on the i7 3770k, this time under three different scenarios:
- Using the stock Intel thermal grease
- Using OCZ’s Freeze Extreme thermal grease
- Using Coollaboratory’s Liquid Pro thermal grease
Each CPU was tested idle and under 100% load first at stock clocks which is 3.5GHz for the i7 3770k, second overclocked to 4.6GHz at 1.2v. The following results show massive differences between Intel’s thermal grease and that of OCZ and Coollaboratory.
Idle conditions are almost negligible in their differences, since at idle the thermal output is so low that the thermal grease doesn’t have much impact. However, under load we can see rather stark differences, at stock configurations the difference between Intel’s thermal paste and OCZ’s is 8 degrees, with a further 3 degrees difference between OCZ and Coollaboratory’s paste, making the total temperature reductions up to 11 degrees.
Overclocked results further exacerbate the problem, with 15 degrees being the difference between Intel’s and OCZ’s thermal grease. The difference between Intel and Coollaboratory’s pastes is 20 degrees.
What’s promising here is that although this doesn’t totally eliminate the problems of Ivy Bridge being hotter than Sandy Bridge it does reduce the margin of difference a lot. Both thermal paste’s used cost around the £4-7 depending on the retailer so for the overclocker and enthusiast this is a relatively cheap fix. However, for the novice user who may find this too complex or the user who doesn’t want to void their warranty this is of no benefit to them. All we can hope is that Intel releases the next revision of Ivy Bridge CPUs with an improved thermal grease to spare end users the hassle, of having to do it themselves and voiding their warranty. It is also worth noting the set-up used to conduct these tests:
- Thermalright SBE Silver Arrow CPU cooler
- ASUS Maximus V Gene motherboard
Given the high quality (and price) of those components, many other users with lower performing CPU coolers and weak(er) VRM boards may find that their temperatures are higher due to needing more volts and possessing less cooling capacity. Still it looks like the average user will be able to shave off around 20% of their overclocked temperatures and 12% off their stock temperatures with some new thermal grease.
On a negative note though, speculation persists that Intel won’t fix this “problem”, because by making Ivy Bridge a less attractive proposition for performance enthusiasts they will encourage a shift towards the more expensive Sandy Bridge-E X79 platform. When considering current overclocked LGA 1155 performance, although Ivy Bridge offers around 5-10% better clock per clock performance, most Sandy Bridge CPUs can clock 5-15% higher and as a result performance ends up equalling to be pretty much the same: i.e a 5GHz 2500K is roughly identical to a 4.6GHz 3570K in performance. Sandy Bridge wins the battle on thermals (perhaps due to deliberate Intel sabotage?) and Ivy Bridge wins on power consumption.
Finally, we would say more testing is definitely still needed to prove these results: anomalies can and do occur, in addition to human error, data and testing manipulation.
Source: Impress PC Watch