The Official Thermal Interface Material thread

[Space Lynx]

It *claimed* to be a 250W+ TDP cooler. -Grabbed it after reading TPU's review on it.
View attachment 339781
At the time, it seemed like one of the 'best value' coolers one could get.

I'm not particularly interested in going to AIO LC route. (Especially, after I just had 2 'shelved' units fail in quick secession.)
However, the TR Assassins are looking very attractive, today.

View attachment 339785

https://www.amazon.com/Thermalright-Peerless-Assassin-120-Cooler/dp/B0C5MBLHD2

yeah I prefer dual tower air coolers myself. Thermal right Frost Commander 140 is the king imo, but the assassin is a close second for sure. @freeagent which thermal right cooler would you recommend him to tame his 5800x3d? also, just use the paste the thermalright cooler comes with, there is nothing wrong with it. best stick with simple, sometimes going fancy just ends up not working.

 

[freeagent]

I have one running under a PS120 EVO. It handles my 58X3D with no trouble at all.

Just my opinion here, but I think old cooler designs cannot handle modern CPUs.

I have pretty much all of the upper end Thermalright coolers, they are impressive.

I also have some old ones that were impressive.. not so much anymore

Edit:

I also use SYY-157, it is cheap, and it is the bestest.. or up there with them.

 

[LabRat 891]

yeah I prefer dual tower air coolers myself. Thermal right Frost Commander 140 is the king imo, but the assassin is a close second for sure. @freeagent which thermal right cooler would you recommend him to tame his 5800x3d? also, just use the paste the thermalright cooler comes with, there is nothing wrong with it. best stick with simple, sometimes going fancy just ends up not working.

The Frost Commander 140 was also 'coming up' as the best air cooler, in my research.
I do need to measure clearance on my 'new' case, though. I'm a bit doubtful a nearly 2 decade old all-alu. Lian-Li has 14cm tower coolers in mind.

edit: Looks like the 'ole Lian Li PC60-derivative has about 6" of clearance for a tower cooler.

The case will not be windowed, and is 'in the white'; would the FC140 black be any tangible improvement over the FC140? (or, just look nicer?)

 

[freeagent]

The case will not be windowed, and is 'in the white'; would the FC140 black be any tangible improvement over the FC140? (or, just look nicer?)

Nah. I have the grey and white FC140 and their performance is identical..

 

[Count von Schwalbe]

. proper 'paste' will flow into the HDT's very uneven base much-better than PTM7950. -making better contact with the 5800X3D's IHS.

Lap the cold plate

adding a thermal shim would act like a coldplate base for the HDT cooler, more evenly cooling the 5800X3D's MCMs

If the cooler base doesn't make contact with the shim it is no better, and if it does it only acts as a thicker IHS with worse thermal conductivity.

 

[freeagent]

Lap the cold plate

No cold plate, just pipes and canyons

 

[Count von Schwalbe]

No cold plate, just pipes and canyons

Dang, you're right. LabRat could still lap it but the gains would be less and there would be risk of taking too much off.

Sounds like a thinner paste might actually be of use. I run a PA-120SE with TF-7 - I can try stress-testing it tomorrow and seeing if it holds up.

Is Prime95 detected and down clocked on 58X3D?

Hadda thought - if you were willing to fabricate a small heat spreader/shim to fit between the heat pipes and the IHS, you could use liquid metal or even solder between the new heat spreader and the heat pipes, dramatically improving heat transfer.

 

[LabRat 891]

Ending up realizing no manner of TIM or cooler mods will 'do'.
The Windale 6 is simply too tall.

Thanks for the input; moved my inquiry to a dedicated thread:

New Air Cooler for 5800X3D (height constrained, retro chassis)

Currently, I have an HDT-type FSP Windale 6 atop my 5800X3D, and regularly bounce off the CPU's thermal limits. Yes, I'm aware I can improve the situation, with better (ducted) airflow and a good paste (instead of PTM7950). Looking for a $30-90,

www.techpowerup.com

 

[Count von Schwalbe]

I run a PA-120SE with TF-7 - I can try stress-testing it tomorrow and seeing if it holds up.

Click to enlarge. I guess it could run faster, but small FFT is pretty rough. It isn't necessarily the most reliable indicator.

 

[freeagent]

I just let p95 run for a little bit as I have to go to work, I can let it run longer when I get home..



I ran Linpack for a couple of iterations and you can see everything gets hit harder by it

 

[lexluthermiester]

While looking for other things in Ebay, I found the the following TIM;

Looked interesting and at only $15 for two 30g tubes, I figured why not. Put it through the standard paces and it performs well. Only 3C on average away from MX6, 4C from MX5 and on par with or better than just about everything else. Damn good deal!

 

[Shrek]

HY610 10g Gold Thermal Grease in the Can-Shenzhen Halnziye Eelctronics Co.Ltd.

Should PC Builders Buy This in BULK? (youtube.com)

 

[TPCEA]

While looking for other things in Ebay, I found the the following TIM;
View attachment 345832
Looked interesting and at only $15 for two 30g tubes, I figured why not. Put it through the standard paces and it performs well. Only 3C on average away from MX6, 4C from MX5 and on par with or better than just about everything else. Damn good deal!

With thermal conductivity of 3.5 W/mK vs Arctic MX-5's 8.5 W/m-K, I'm curious how accurately manufacturers rate their TIMs.

 

[lexluthermiester]

HY610 10g Gold Thermal Grease in the Can-Shenzhen Halnziye Eelctronics Co.Ltd.

Really doesn't look "Gold" and I promise you it has no actual Gold in it. Kinda looks more Brass-like. Still, performs great, it's easy to apply and inexpensive! It's a winner so far in my book.

With thermal conductivity of 3.5 W/mK vs Arctic MX-5's 8.5 W/m-K, I'm curious how accurately manufacturers rate their TIMs.

I never take those numbers seriously. While the equation is the same, how to measure and calculate it isn't. There is no standardized way to do it and as such, those stated numbers are completely meaningless. We can only ever know the performance by testing it ourselves.

Should PC Builders Buy This in BULK? (youtube.com)

I don't think that's the same stuff. Product number is similar, but that video is 7 years old. Might not apply.
EDIT1: Or it is the same. Just started watching and saw the check mark...
EDIT2: I trust my methodology of testing far more than the nonsense Linus displays in that video. Typical LTT slap-happy carelessness.

 

[Shrek]

Looking forward to your results.

If you ever have time GD900 is about $2.50 for 30g (with postage)
30g High Performance Heat Sink Compound Grease Paste GD900 New Termica` | eBay

 

[lexluthermiester]

Looking forward to your results.

See above post. Or were you suggesting a fully documented testing session?

 

[Shrek]

Ah, missed that 3°C hotter; good stuff.

 

[lexluthermiester]

Ah, missed that 3°C hotter; good stuff.

Yeah, MX5 is still the top performer of the TIMs I have here currently. MX6 closely follows, then Thermalright's TF7, ID Cooling's Frost X25 and on down. Here's the interesting fact I have discovered: Of the TIMs I have in my personal stock, all but one performs within 4 to 5 % of each other. The remaining one TIM is 8% out. Ditched Noctua's TIM's as both NT-H1 and NT-H2 seem to dry out and loose performance a few months after application. I had some KPx, which is another top performer, but ran out of it and haven't bought more. Likely will soon. I like KPx. It's A LOT like MX5.

 

[xtreemchaos]

i can put a plus for GD900 when i first started using it i sorta realised that most other paste is a waste of money yes some are better but only by a few deg may be 2 or 3 c but cost 10 times as much but be aware that there copys "fakes" of it about now .

 

[Sithaer]

i can put a plus for GD900 when i first started using it i sorta realised that most other paste is a waste of money yes some are better but only by a few deg may be 2 or c but cost 10 times as much but be aware that there copys "fakes" of it about now .

Yep, my bro bought the first/original batch of GD900 from Ali before it became more known and we used that on various hardware with no issues.
Actually the performance/texture of the paste was in the same ballpark of the MX 4 we used before for years, I've had 2 GPUs repasted with that GD900 including my 3060 Ti for ~6 months with no issues/it was still keeping the same temps and it wasn't dried out when I took it apart to repaste with my current MX 6.

Tbh I've only bought this MX 6 out of curiosity hence only the 4g tube, luckily in my case it did not degrade yet.
I also agree with Lex, I've used MX 4/6/GD900/whatever my ID cooler came with and the temperature diff was minimal with the level of hardware I/my bro was using.

 

[TPCEA]

i can put a plus for GD900 when i first started using it i sorta realised that most other paste is a waste of money yes some are better but only by a few deg may be 2 or c but cost 10 times as much but be aware that there copys "fakes" of it about now .

Who actually makes it, so I can avoid the fakes?

 

[xtreemchaos]

the fakes from what ive seen is lighter in colour all most white and the real stuff is gray/blue. i brought 3 30g tubes id say 5 years ago and im on my second tube. when i do rigg repairs i give the customer a choice between mx4 for £5 or GD900 for free and when i tell them the only diff is the price most go for the GD. the lable's look the same so do the tubes. the web ad is www.ourGD.net on my real stuff. heres a youtube vid for what its worth.
The Best Thermal Paste for PC CPU, GPU and Laptop (youtube.com)

and heres a pc of the real stuff.

 

[FreedomEclipse]

Apparently GD900 suffers from pump out

 

[xtreemchaos]

Apparently GD900 suffers from pump out

mine don't, maybe if its really cold eh . comes out the tube smooth and says under the cooler nice aswell may be if too much is use it would.

 

[Shrek]

i can put a plus for GD900 when i first started using it i sorta realised that most other paste is a waste of money yes some are better but only by a few deg may be 2 or 3 c but cost 10 times as much but be aware that there copys "fakes" of it about now .

Good thing I got four lots of 30g some time back.