Ivy Bridge and changing the Thermal Interface Material

Now that you know what the aim of all this testing is, we’d now like to elaborate on our testing method so we can be clear to you how the results were achieved and if you want to replicate them then you can attempt to do so following a similar method.

We have tried to target this at the widest possible audience by bringing the financial “hit” of the components as low as possible. Using the i5 3570K instead of the i7 3770K this is a much more affordable opportunity. We’ve also opted for air cooling instead of water cooling or LN2 as this also makes things a heck of a lot cheaper. We won’t go any more into the test system as we will detail that on the next page.

Using a selection of 5 thermal pastes we will first test the i5 3570K for its temperatures with a stock IHS configuration.We will test the processors maximum load temperatures using 10 minutes of Prime95 for each thermal paste . This testing will be conducted at stock speeds of 3.4GHz (which can go to 3.8GHz through Turbo) and overclocked speeds of 4.5GHz. The reason we have not included idle temperatures is because Intel’s Speed Step technology means an idle processor produces almost no heat at all, so all thermal pastes perform almost identically.

So we will have these results for each thermal paste in the stock IHS configuration:

• 3.4GHz (3.8GHz Turbo) maximum load temperatures
• 4.5GHz maximum load temperatures

Next we have to remove the IHS. This will be done by using a flat single sided razor to gently cut under each corner of the IHS by a few millimetres, the IHS should then become loose and can be fully removed. Upon removal the stock Intel applied TIM will be visible. We will then remove this using an high alcohol-content liquid. Upon removal we will then go through the same testing process as on the stock IHS except this time we apply the thermal pastes above and below the IHS each time. We also intend to drop any thermal pastes that fall behind the rest in terms of performance.

We again will have the following results for each thermal paste:

• 3.4GHz (3.8GHz Turbo) maximum load temperatures
• 4.5GHz maximum load temperatures

Note we use Delta temperatures, that is Load temperature minus the ambient temperature. This allows us to more accurately account for the effects of changing room temperatures.