Delidding – Kyrosheet vs PTM7950 as TIM

Hey folks, hope everyone is keeping well 🙂

Continuing my adventures on seeing how different things that aren’t LM perform as TIM for my 12900k (between the Die & IHS), I grabbed some Kryosheet recently as it was available locally (which is a surprise in itself) and at a reasonable price as well, and thought I’d post this up since I was previously running PTM7950 and I’ve seen a fair few requests for this, but also trying to establish exactly what PTM7950 is good to run on (and when not to), and the same for Kryosheet.

So without further ado – the results: I won’t throw up the PTM pictures here since I think they already exist somewhere else on the sub, but the important information is that with PTM7950, I could get about 150w-160w of heat through the CPU before it thermal throttled. Clocks wise it could only manage about \~4.6ghz due to the above (with a -0.110v undervolt). So while PTM is pretty solid for many things, unfortunately it simply hasn’t got the thermal transfer ability for this sort of concentrated application. It is worth noting that there is a bit of a delta on the cores, where one core is about 5c higher than the rest with both the PTM and Kyrosheet. I think in both cases if I changed the clocks on that core, then the overall performance may increase but I decided to leave it as is for the moment. This is also on a 240mm AIO, and I used PTM7950 between the IHS and Coldplate for both tests, so all the gains/changes are purely from the TIM application – there’s still potentially more on the table if I did change the PTM between the IHS and Coldplate (unsure if that’s the limiting factor at this stage).

Now for the Kyrosheet – In my application, seeing as it is Graphite with the crazy horizontal (for simplicity’s sake) heat transfer, I decided to make my sheet IHS sized rather than die sized, and actually used some Kapton tape (Polymide tape for the non-branded stuff) to hold it onto the IHS, in the hopes that it may make better IHS contact and maybe wick away extra heat. Whether it worked, I have no idea haha. Post Kyrosheet, I’m now getting stock clock speeds (4.9Ghz P-Cores, 3.7Ghz E-Cores with all cores active) initially at about 90-92c on that same hottest core (P-Core 7) and about \~195w. In fact I had forgotten about my PL1 setting, so had to change that as well to even allow it to get anywhere near high enough! So already an improvement.

The next day, I booted it up and benched it again – in fact to actually test the Kyrosheet, I had to apply a mild overclock (what a shame, so regrettable haha) just to 5.0ghz on the P-cores, and 3.9ghz on the E-Cores. Same hottest core (P-Core 7) was throttling at 100c, however power draw was up to \~215w. There’s still about a 5c delta between the average P-Cores and that one hotter core, but the hottest E-Core was only 84c, so plenty of headroom there if I did want to pump those up, but I couldn’t be bothered messing with my undervolt so I left them as is.

So, would I recommend the Kyrosheet? From my testing I’d say definitely! It is however expensive, and there are considerations as to whether I’d use it in a particular application – namely whether you have the thermal density to actually need it, as if you’re only pushing 100w say through a ‘regular’ sized CPU die, you can probably get away with PTM7950 pretty happily I think. As mentioned the limit is about 150w-160w, but I imagine above about 135w that differential will start increasing and that’s when I’d be considering Kyrosheet. In theory it should be better in all applications due to the better thermal transfer, but whether that’s noticeable is another thing as well. Second is the finicky nature of installing – PTM is easy enough once cooled to apply and will literally melt the excess off the sides once running (I’ve got pictures of this after it’s happened if anyone wanted), whereas Kyrosheet will move around and basically hates you. The sheet isn’t super fragile, still delicate, you also don’t want to bend it over itself, and considering how easily everything slides (even just clamping the IHS in the socket the IHS would slide around considerably even with me holding it down), I’d actually consider cutting the Kyrosheet slightly larger than it needs to be and taping it in place to hold it still during installation. Also you do have to be careful with this since it’s conductive (PTM is not), but Polymide tape is excellent for isolating things and also holding the sheet down I’ve found.

So those are my findings! I’ve put the images in the Imgur gallery here ([https://imgur.com/a/kPWeazl](https://imgur.com/a/kPWeazl)) since Reddit makes them super large and in-the-way. But hopefully this is of use to someone! Now to convince myself that I don’t need to do the GPU haha, because I don’t need to but want to 😀