TIM under Fermi's cap..

[Hauk]

Most of us wouldn't try to remove a heatspreder from an expensive GPU. Someone did and here's what's under Fermi's:

Thanks to Largon at TPU for these insightful pics. Largon describes the adhesive holding the IHS on as a hard epoxy that's brittle enough to give way with a little force. He also notes a blade alone is not enough to scrape it off.

What else do we see.. a TIM that appears to be the same stuff as what they use topside. I'm usually up for replacing stock TIM, but I'll not mess with this one. There it is though, another interface..

 

[railven]

Most of us wouldn't try to remove a heatspreder from an expensive GPU. Someone did and here's what's under Fermi's:

Thanks to Largon at TPU for these insightful pics. Largon describes the adhesive holding the IHS on as a hard epoxy that's brittle enough to give way with a little force. He also notes a blade alone is not enough to scrape it off.

What else do we see.. a TIM that appears to be the same stuff as what they use topside. I'm usually up for replacing stock TIM, but I'll not mess with this one. There it is though, another interface..

The pictures won't load (firewall at work) so if I read your post correctly:

There is a TIM on the dies between the heatspreader? Is this common for any kind of packaging?

Just curious if these are the new steps EE are using to keep these modern day monster chips cooler.

 

[tincart]

That is one monster chip underneath.

 

[Worthington]

Here is an excellent step by step guide (w/pics) if you're interested in replacing the TIM on the 480. Depending on your vendor it's covered under the warranty. Some people see a noticable difference afterwards, some not so much.

http://www.evga.com/forums/tm.aspx?m=310095

 

[Hauk]

Here is an excellent step by step guide (w/pics) if you're interested in replacing the TIM on the 480. Depending on your vendor it's covered under the warranty. Some people see a noticable difference afterwards, some not so much.

http://www.evga.com/forums/tm.aspx?m=310095

Yep, similar for the 470. That details replacing TIM making contact with the cooler/heatsink though. Pictured is the TIM under the heatspreader. I always assumed the die contacted the heatspreader directly. Not so.. there's TIM underneath. Begs to question, could a few degrees be shaved by replacing this TIM? Not worth the risk IMO..

 

[lopri]

So was s/he successfully able to remove the IHS without damaging anything? That's an amazing job. Thank you for the pics I don't think I'd ever seen the actual (physical) die of Fermi.

 

[lopri]

There is a TIM on the dies between the heatspreader? Is this common for any kind of packaging?

I believe so. And IHS removal was sort of popular back in the A64 days. You can see TIM between die and IHS.

http://www.youtube.com/watch?v=j8WWb0aCFvk

 

[Hauk]

So was s/he successfully able to remove the IHS without damaging anything? That's an amazing job. Thank you for the pics I don't think I'd ever seen the actual (physical) die of Fermi.

No damage done. A mod said he's old school, been tearing stuff apart and posting for years. He mentioned GT200 was same way, with TIM under the IHS.

 

[EarthwormJim]

The pictures won't load (firewall at work) so if I read your post correctly:

There is a TIM on the dies between the heatspreader? Is this common for any kind of packaging?

Just curious if these are the new steps EE are using to keep these modern day monster chips cooler.

There's always TIM between a heat spreader and the core. It's not always a paste though or epoxy like substance.

No piece of metal is perfectly flat or smooth, and there will always be a small air gap so TIM is necessary.

 

[GodisanAtheist]

There's always TIM between a heat spreader and the core. It's not always a paste though or epoxy like substance.

No piece of metal is perfectly flat or smooth, and there will always be a small air gap so TIM is necessary.

-Yeah, not sure what he was expecting to find. Leprechauns maybe? Elfs?

Nevertheless, more power to him for having the fuzzy mansacks to take that thing apart. Anyone know if getting the heat spreader off a G80 core is any different?

 

[Acanthus]

I'd love to do this to try direct watercooling the GPU, like people do with CPUs over at XS every now and again.

 

[cbn]

I'd love to do this to try direct watercooling the GPU, like people do with CPUs over at XS every now and again.

Do you happen to know a good thread had those results?

P.S. IBM is using chip level watercooling with aquasar, but they are using warm water instead of chilled.

 

[Acanthus]

Do you happen to know a good thread had those results?

P.S. IBM is using chip level watercooling with
aquasar, but they are using warm water instead of chilled.

Very nice, letting the water get warm in a supercomputer makes sense if you are going to repurpose the hot water for heating rooms.

Genius move on their part.

 

[faxon]

i thought they normally used solder under the IHS because it's got better heat conductivity. am i wrong in this assumption?? i know intel uses solder, i have seen some intel cpu deliddings where you could see some of the remaining solder on the die as well

 

[EarthwormJim]

i thought they normally used solder under the IHS because it's got better heat conductivity. am i wrong in this assumption?? i know intel uses solder, i have seen some intel cpu deliddings where you could see some of the remaining solder on the die as well

I think the majority of Intel and AMD CPUs use solder. Lower end parts do not.

 

[Hauk]

I'd love to do this to try direct watercooling the GPU, like people do with CPUs over at XS every now and again.

I think that's what Largon was doing with his..