Discussion Thermal Paste versus Graphite Thermal Pads

[Golgatha]

I recently tried out a graphite thermal pad for a new Intel 9400F build and it worked quite amazingly well. Temps with a Zalman 9500 cooler are barely above ambient and around 55°C running Prime 95 torture. Similar performance to my normal AS5 paste, but without the mess. Similar results with an old Xeon X3440 system.

This is what I used - https://www.amazon.com/gp/product/B07CK9SHZG

30x30 for the 9400F and an old Xeon X3440

Going to try a 40x40 for my old X5820 system. Anyone else used these with as good of results? I'm also wondering if there's some thermal limit to them, as the X5820 is a 140w CPU. If anyone has used one of these with a similar TDP, please let me know how it went. TYIA!

 

[mindless1]

Seems kind of expensive unless you're going for extreme o'c. I haven't found the need to pay for exotic paste or pads ever since chips got heat spreaders on them, so my per CPU goop cost is maybe 15 cents?

 

[Bavor]

Seems kind of expensive unless you're going for extreme o'c. I haven't found the need to pay for exotic paste or pads ever since chips got heat spreaders on them, so my per CPU goop cost is maybe 15 cents?

Supposedly the pads transfer heat about three times better than some of the better thermal compounds on the market and result in lower CPU temperatures. I've seen others say the pads are delicate and may rip if you remove the heat sink or water block.

 

[mindless1]

^ Unless I am doing an overclocking contest, what do I care if my CPU is 3C cooler as long as it's stable? I have never ruined a CPU from long term heat, they always get replaced due to something better faster coming along.

If I put a slightly better thermal interface down, it just delays a bit how long till the heatsink fan hits each ramp up in speed, so the amount of noise and dust buildup is reduced a little, except that I can't hear my CPU 'sink fan as it is, and my case has a filtered intake so dust doesn't build up before I'm upgrading again.

I was very much an advocate of good CPU - heatsink interfaces back in the day. I even lapped an Athlon XP and heatsink to the point where it stayed cool enough with NO heatsink compound, just a dry joint. What did I learn? It's not worth the bother, the only thing worth considering for a system you regularly use rather than an experiment, is that it isn't just a silicone oil and zinc oxide paste that separates and drys out leaving behind crusty islands with low thermal conductivity. Thermal paste has gone way beyond that without having to spend much to achieve it.

In context, if I can run Prime95 at 55C, or it would be 62C with any random goop, I didn't gain anything, just lost the cost of the exotic pad, wasted that money.

On the other hand, all this attention to details is a hobby and there are plenty of ways to spend more money on a hobby. At least it does "something", but the something it does isn't important if the temp isn't high enough that it's bordering on the threshold for instability... and it might be if you are doing extreme overclocking, but I don't need that last 100-200MHz that bad. There will always be something faster a ~year later, so it is chasing a ghost.

 

[Furious_Styles]

^ Unless I am doing an overclocking contest, what do I care if my CPU is 3C cooler as long as it's stable? I have never ruined a CPU from long term heat, they always get replaced due to something better faster coming along.

If I put a slightly better thermal interface down, it just delays a bit how long till the heatsink fan hits each ramp up in speed, so the amount of noise and dust buildup is reduced a little, except that I can't hear my CPU 'sink fan as it is, and my case has a filtered intake so dust doesn't build up before I'm upgrading again.

I was very much an advocate of good CPU - heatsink interfaces back in the day. I even lapped an Athlon XP and heatsink to the point where it stayed cool enough with NO heatsink compound, just a dry joint. What did I learn? It's not worth the bother, the only thing worth considering for a system you regularly use rather than an experiment, is that it isn't just a silicone oil and zinc oxide paste that separates and drys out leaving behind crusty islands with low thermal conductivity. Thermal paste has gone way beyond that without having to spend much to achieve it.

In context, if I can run Prime95 at 55C, or it would be 62C with any random goop, I didn't gain anything, just lost the cost of the exotic pad, wasted that money.

On the other hand, all this attention to details is a hobby and there are plenty of ways to spend more money on a hobby. At least it does "something", but the something it does isn't important if the temp isn't high enough that it's bordering on the threshold for instability... and it might be if you are doing extreme overclocking, but I don't need that last 100-200MHz that bad. There will always be something faster a ~year later, so it is chasing a ghost.

No mess is nice, as someone who repairs mobo it's simpler and creates less waste. If anyone wants to get one for a long term build I can only recommend them.

 

[UsandThem]

Those pads are a little pricey, but if they held up for long periods of time (longer than say decent quality paste), and provided similar to slightly better temperatures, I would use one.

 

[WelshBloke]

I've got one on my CPU. It works fine. I'm not sure that theres any functional difference between it and any other thermal material with regards to operating temperature, fan noise or overclock.
I got it because;
A) I was curious about them.
B) Theres no mess and its easier to apply and get even than paste.
C) The cost is pretty negligible when you factor in the motherboard and CPU that went with it!

I would absolutely use them again.

 

[VirtualLarry]

What did I learn? It's not worth the bother, the only thing worth considering for a system you regularly use rather than an experiment, is that it isn't just a silicone oil and zinc oxide paste that separates and drys out leaving behind crusty islands with low thermal conductivity. Thermal paste has gone way beyond that without having to spend much to achieve it.

This, entirely. Unless you're an "extreme over-clocker" (and 9400F isn't even overclockable!), the best "lesson" for TIM, is to "use something better than 'white paste'", and then don't worry about it. Spread method and overall technique often matter more than the particular paste used, once you get above the "white paste" level.

 

[Golgatha]

Thanks all for the comments. I figure the $10 I spent will allow me to not ever worry about the thermal paste drying out. I'm putting these in systems I don't plan to mess with for a long period of time, say 3+ years. Also, I can conceivably reuse them later. As others have said, it's a pretty small percentage money wise compared to the total cost of each system as well.

 

[zir_blazer]

^ Unless I am doing an overclocking contest, what do I care if my CPU is 3C cooler as long as it's stable? I have never ruined a CPU from long term heat, they always get replaced due to something better faster coming along.

If I put a slightly better thermal interface down, it just delays a bit how long till the heatsink fan hits each ramp up in speed, so the amount of noise and dust buildup is reduced a little, except that I can't hear my CPU 'sink fan as it is, and my case has a filtered intake so dust doesn't build up before I'm upgrading again.

[...]

In context, if I can run Prime95 at 55C, or it would be 62C with any random goop, I didn't gain anything, just lost the cost of the exotic pad, wasted that money.

Wasn't there a hugely referenced Thread by Idoncare where he showed how temperature affects power consumption (All the photos are broken, the text version is at least parseable)? If the difference between 60°C and 55°C is 2-3W on power consumption, then you just have to calculate how many months it would take for the Thermal Pad to pay for itself.
Assuming that you're running 24/7 and saving 3 W-h, you're saving 1 Kw-h per 14 days. How much does a Kw-h cost? How much time it would take?

 

[Golgatha]

Wasn't there a hugely referenced Thread by Idoncare where he showed how temperature affects power consumption (All the photos are broken, the text version is at least parseable)? If the difference between 60°C and 55°C is 2-3W on power consumption, then you just have to calculate how many months it would take for the Thermal Pad to pay for itself.
Assuming that you're running 24/7 and saving 3 W-h, you're saving 1 Kw-h per 14 days. How much does a Kw-h cost? How much time it would take?

Interesting. I'd potentially save about $0.30 a month, so about 3 years for a break even point.

 

[Hitman928]

Wasn't there a hugely referenced Thread by Idoncare where he showed how temperature affects power consumption (All the photos are broken, the text version is at least parseable)? If the difference between 60°C and 55°C is 2-3W on power consumption, then you just have to calculate how many months it would take for the Thermal Pad to pay for itself.
Assuming that you're running 24/7 and saving 3 W-h, you're saving 1 Kw-h per 14 days. How much does a Kw-h cost? How much time it would take?

Interesting. I'd potentially save about $0.30 a month, so about 3 years for a break even point.

Unfortunately Idontcare didn't isolate temperature in his raw data post, otherwise I could re-plot this for you. It would also be very interesting to see how his findings would change with finfets. He mentions that he believes the bulk of the power consumption increase happens with increased temps because of the increase in static power consumption. Traditionally dynamic power consumption dominated the power use of CPUs but as the nodes got smaller and smaller, static power consumption became a bigger and bigger share of the power used by a CPU. He is testing a Sandy Bridge CPU for his tests which were at 32 nm. At 32 nm the static power consumption was already becoming a large problem for planar, bulk processes and it become pretty much untenable at 22 nm. This was a huge drive to get finfets developed as they greatly improve the static power consumption of FETs which is what allowed for continued high performance digital processors at 22 nm and beyond.

Edit: Nevermind, I do see where he isolated temperature at certain frequencies and voltages in the raw data, I'll see if I can re-plot those results if I have time today.

 

[Hitman928]

From IDC's raw data. Maybe over the holiday break I'll find some time to repeat the experiment with a finfet CPU.

 

[lobz]

I think, for inexperienced users the $10 is literally nothing compared to the problems they can get by not applying the TIM correctly, especially because they might not even be aware of it.

I used TIM for my current build (still rocking a 4790K) but I'm seriously considering using a pad for the next one, mainly because there are not many people out there, that are lazier than me

 

[Ranulf]

I have a pad so I don't have to mess with cleaning cpu and heatsink when just doing testing or bios upgrades etc.. I only have to worry about cleaning paste off a heatsink I need to use that still has some on it from the factory or previous use.

For long term builds I still use paste.

 

[thor23]

From what I've heard the pads just last longer than thermal paste. Compared to freshly applied paste they perform worse by 2-3c, but after 5 years the pads will still perform like new and the paste loses effectiveness.
Doesn't really matter for a non oc'd pushed to the limit system they can run stock heatsink/paste forever.

 

[DAPUNISHER]

No more n00bs pulling the CPU out when they take the cooler off, and with AMD, bending pins in the process. No more mess, no more youtube comments giving you guff for how you apply your t.i.m. And if you swap CPUs on the same test setup like some of us do, they are a nice convenience. Phil's Hardware gave me the idea.

 

[WelshBloke]

No more n00bs pulling the CPU out when they take the cooler off, and with AMD, bending pins in the process.

I have to say that did come into my decision making! I swapped out coolers a few times when I first got this CPU and the pad was just easier.

 

[chrisjames61]

Using a thermal pad over paste is like fixing something that isn't broken.

 

[WelshBloke]

Using a thermal pad over paste is like fixing something that isn't broken.

You get the same result in an easier cleaner way. That seems like an improvement to me!

 

[Charlie22911]

I wonder if there be a benefit to using these in place of traditional thermal pads for VRMs? I’m severely thermally limited by VRM temps right now...

 

[thor23]

Yes they're much superior to normal thermal pads, almost as good as thermal paste.

 

[Bavor]

From what I've heard the pads just last longer than thermal paste. Compared to freshly applied paste they perform worse by 2-3c, but after 5 years the pads will still perform like new and the paste loses effectiveness.
Doesn't really matter for a non oc'd pushed to the limit system they can run stock heatsink/paste forever.

I've heard the opposite about the graphite thermal pads. Users are claiming 2C to 3C better temperatures than they obtained with Thermal Grizzly Kryonaut. Maybe it depends on the CPU.

 

[DrMrLordX]

I've heard the opposite about the graphite thermal pads. Users are claiming 2C to 3C better temperatures than they obtained with Thermal Grizzly Kryonaut. Maybe it depends on the CPU.

That surprises me. The pads on Radeon VII were not very popular with those of us who bought them, though those pads may be inferior to the ones we're discussing here. Those pads were definitely not reusable. They fragment easily and are kind of a pain to remove.

 

[Bavor]

That surprises me. The pads on Radeon VII were not very popular with those of us who bought them, though those pads may be inferior to the ones we're discussing here. Those pads were definitely not reusable. They fragment easily and are kind of a pain to remove.

The issue on the Radeon VII was the mounting pressure and sometimes the uneven die surface due to the different thickness die and HBM2 chips.