Quick Cooling Tip For Ivy

[Cecil]

Been fighting the heat with Ivy since I got it like everyone else. Just using my H100 right now for initial testing. I was seeing how far I could get with no HT, and at 4.8Ghz 1.385Vcore, I was hitting 93C
I realised one core was way down at 77C though, so I figured Id better check the TIM.

I had done a simple dot in the middle, and boy was that a bad idea. Got out my spreader that came with a tube of TX-4. Spread out a nice thin, very even layer, the exact size as the contact area of the IHS. Now, hottest core is 68C, coolest is 62C.

So, quick tip is, dont do a dot in the middle. Spread out a very thin, and very even layer on your heatsink. Press/clamp/screw it down on the CPU first if its new to get that indentation of the IHS on the heatsink so you know what area exactly needs to be covered.

Remember that the Ivy ide is a long rectangle shape in the center of the IHS, so a dot wont get to the ends. A line might work, but you want as much contact as possible with the full IHS. Dont leave any not making good contact.

[Stridulent]

Originally Posted by Cecil

Been fighting the heat with Ivy since I got it like everyone else. Just using my H100 right now for initial testing. I was seeing how far I could get with no HT, and at 4.8Ghz 1.385Vcore, I was hitting 93C
I realised one core was way down at 77C though, so I figured Id better check the TIM.

I had done a simple dot in the middle, and boy was that a bad idea. Got out my spreader that came with a tube of TX-4. Spread out a nice thin, very even layer, the exact size as the contact area of the IHS. Now, hottest core is 68C, coolest is 62C.

So, quick tip is, dont do a dot in the middle. Spread out a very thin, and very even layer on your heatsink. Press/clamp/screw it down on the CPU first if its new to get that indentation of the IHS on the heatsink so you know what area exactly needs to be covered.

Remember that the Ivy ide is a long rectangle shape in the center of the IHS, so a dot wont get to the ends. A line might work, but you want as much contact as possible with the full IHS. Dont leave any not making good contact.

I experienced very similar results when I did the "line" method. I reapplied using an "X" instead of a line and my temps were about 5 degrees lower and more consistent with eachother. Might try manually spreading it next time I redo it.

[hostage67]

Originally Posted by Cecil

Been fighting the heat with Ivy since I got it like everyone else. Just using my H100 right now for initial testing. I was seeing how far I could get with no HT, and at 4.8Ghz 1.385Vcore, I was hitting 93C
I realised one core was way down at 77C though, so I figured Id better check the TIM.

I had done a simple dot in the middle, and boy was that a bad idea. Got out my spreader that came with a tube of TX-4. Spread out a nice thin, very even layer, the exact size as the contact area of the IHS. Now, hottest core is 68C, coolest is 62C.

So, quick tip is, dont do a dot in the middle. Spread out a very thin, and very even layer on your heatsink. Press/clamp/screw it down on the CPU first if its new to get that indentation of the IHS on the heatsink so you know what area exactly needs to be covered.

Remember that the Ivy ide is a long rectangle shape in the center of the IHS, so a dot wont get to the ends. A line might work, but you want as much contact as possible with the full IHS. Dont leave any not making good contact.

Ya, that method went out with the introduction of multi-core CPU's. I have now started using the spread method where you use a business card to spread a thin layer all across the CPU and it seems to be working well for me.

[ShrimpBrime]

Originally Posted by Cecil

Spread out a nice thin, very even layer, the exact size as the contact area of the IHS. Now, hottest core is 68C, coolest is 62C.

Man I always spread. Every Cpu. Especially the hot ones.

So Cecil's thread is a nice conclusion to this tutorial thread IMO http://forums.extremeoverclocking.co...d.php?t=343346

[Playap0wnr]

What about a cross design since you can't tell where the rectangle die is?

[DrGuns4Hands]

Originally Posted by Playap0wnr

What about a cross design since you can't tell where the rectangle die is?

Why wouldn't you cover the whole thing? then you know you got it.

Also if anyone has the guts to delid the CPU I hear Ivy's TIM between the die and case is pretty bad, and you can get another 10c+. http://www.eteknix.com/news/ivy-brid...ems-after-all/

[Carozzeria]

+1 to Shrimpbrime.

I've never understood the dot method. If the CPU comes with an integrated heat spreader it should be fully exploited.

[hostage67]

Originally Posted by Carozzeria

+1 to Shrimpbrime.

I've never understood the dot method. If the CPU comes with an integrated heat spreader it should be fully exploited.

The dot method was a way to spread the paste without bubbles. If you spread the paste manually and then place the heatsink then you will most likely end up with some bubbles, which means it won't cool as well.

When it was a single core processor the dot method was the most effective. You simply place a dot in the center and when you mount the heatsink it will spread the paste across the IHS without creating bubbles.

Now that we are working with multiple cores however, it has become much more challenging to get the dot method to work. First off you run the risk of missing part of core if you don't know exactly where they are placed on the CPU, and secondly now that you have multiple dots they are prone to create bubbles where they meet anyways. So given the risk of exposure and the fact that you are getting bubbles anyways, spreading it manually eliminates the risk of exposure which is the more critical issue to begin with.

[leekirlew]

i always spread it myself, nice and smooth across the whole ihs, seems to work best for me

[ShrimpBrime]

Originally Posted by hostage67

The dot method was a way to spread the paste without bubbles. If you spread the paste manually and then place the heatsink then you will most likely end up with some bubbles, which means it won't cool as well.

When it was a single core processor the dot method was the most effective. You simply place a dot in the center and when you mount the heatsink it will spread the paste across the IHS without creating bubbles.

Now that we are working with multiple cores however, it has become much more challenging to get the dot method to work. First off you run the risk of missing part of core if you don't know exactly where they are placed on the CPU, and secondly now that you have multiple dots they are prone to create bubbles where they meet anyways. So given the risk of exposure and the fact that you are getting bubbles anyways, spreading it manually eliminates the risk of exposure which is the more critical issue to begin with.

Now spreading on AMD FX-4100 with the stock cooler is pointless. The center of the stock HSF they give only touches and the IHS lpate is actually exposed to the air, meaning only the copper core touches the IHS plate and the aluminum part (fins) doesn't even touch.

So spreading and dot techniques may vary depending on the cooler. The heat pipe coolers that actually place heat pipes right to the Cpu require the actual heat sink to be TIM'ed then you nmay as well spread on the Cpu as well.

So many variables to say who's doing what.

A good to better than average cooler should touch the entire surface area of the Cpu IHS plate. If it doesn't, it will not remove the heat as effectively. Most water blocks have really good coverage.

Additional Comment:

Oh and bubbles while spreading is rare imo. Usually the Cpu feels like it's gonna rip out of the socket cause the darn thing is suction cup'ed like. Come to think of it, I've ripped my fare share of cpu's from motherboards lol.