The Official Thermal Interface Material thread

[8BitZ80]

Here's a thermal paste that should be avoided: Unick Silicone Heat Transfer Compound.

I bought some to try because I had a tube around 15 years ago when I knew nothing about computers. I didn't know how to compare thermal paste performance and so I never really knew if it was good or not. The main benefit is that you get 150 grams for around $15 or so.

Well, now I can safely say that it's junk. It runs about 8C hotter than an average thermal paste. Physically this stuff is very similar to sunscreen, except with no smell. It's very difficult to clean because it's hard to see it on paper towel and it smears itself into every little crevice that it can find. Not recommended at all.

If you want GOOD bulk thermal paste, I'm still standing by GD900 from Aliexpress as the best option. The cost-per-volume is similar but it performs far better, more like a standard Arctic MX-4 paste. Much easier to apply and clean off components as well.

 

[8BitZ80]

I did a performance shootout between Noctua NT-H1 (AU$4.5/gram) and GD900 (AU$0.14/gram) which yielded absolutely identical results. Deep down I wanted NT-H1 to win because then it would at least be somewhat justifiable to spend 30 times more on a retail thermal paste. I gave it every opportunity to inch even ONE degree ahead but it refused.

Testbench is an overclocked Pentium G3258 with a dual heatpipe tower cooler running Molex-powered fans. Ambients were 28degC on each run, verified at start and finish. I subjected each thermal grease to 30 minutes at 100% load using MSI Kombustor and took screenshots for each one on completion.

I'm going to be re-visiting this test in the future (hopefully with more test subjects like Arctic MX-4), but with the tools I've got on hand right now I'm struggling to find any performance differences at all. You actually have to study the images carefully to find any differences, but you can definitely see the screenshot that I took for NT-H1 first.

I also did a preliminary test earlier and that resulted in identical maximum temperatures, so there's reason to believe this isn't a fluke result.

Noctua NT-H1:

GD900:

 

[Lindatje]

I use LM on my CPU and GPU, nothing is better.

 

[Nater]

I did a performance shootout between Noctua NT-H1 (AU$4.5/gram) and GD900 (AU$0.14/gram) which yielded absolutely identical results. Deep down I wanted NT-H1 to win because then it would at least be somewhat justifiable to spend 30 times more on a retail thermal paste. I gave it every opportunity to inch even ONE degree ahead but it refused.

Testbench is an overclocked Pentium G3258 with a dual heatpipe tower cooler running Molex-powered fans. Ambients were 28degC on each run, verified at start and finish. I subjected each thermal grease to 30 minutes at 100% load using MSI Kombustor and took screenshots for each one on completion.

I'm going to be re-visiting this test in the future (hopefully with more test subjects like Arctic MX-4), but with the tools I've got on hand right now I'm struggling to find any performance differences at all. You actually have to study the images carefully to find any differences, but you can definitely see the screenshot that I took for NT-H1 first.

I also did a preliminary test earlier and that resulted in identical maximum temperatures, so there's reason to believe this isn't a fluke result.

Noctua NT-H1:

View attachment 196933

GD900:

View attachment 196932

I went to look at Newegg.com here in the US....and 10g of Noctua NT-H1 is $14.95, while 10x 1g tubes of GD900 is $16.29...yet you can get a 30g tube of it for $12.99.

???

I've never heard of GD900 I guess.

 

[Shrek]

Well, now I can safely say that it's junk. It runs about 8C hotter than an average thermal paste.

8C is a lot, but I am not sure it is worth chasing 2C; I worry about longevity and the dangers of liquid metal.

Gallium Induced Failure of a Heat Sink - YouTube

 

[8BitZ80]

I went to look at Newegg.com here in the US....and 10g of Noctua NT-H1 is $14.95, while 10x 1g tubes of GD900 is $16.29...yet you can get a 30g tube of it for $12.99.

???

I've never heard of GD900 I guess.

Yeah I don't know what the deal is with those resellers, but the 30g tubes are sold through AliExpress (GD900 listings). It's so cheap there that they're practically giving it away. Right now they're US$2.60 each. Last time I ordered a couple in March it took three weeks to get them and the new batch performs just as good as the first one I got, so it seems like it performs consistently.

8C is a lot, but I am not sure it is worth chasing 2C; I worry about longevity and the dangers of liquid metal.

(1) Gallium Induced Failure of a Heat Sink - Bing video

Yeah when it's 8 degrees hotter then it's not worth messing with. Pretty much every other thermal paste on the market does better than that stuff.

I'm thinking of doing some testing with my R9 270X to get some more accurate readings on these thermal pastes. It generates a good amount of heat and you can easily run the fans at 100% on a graphics card, so it might be a good candidate for it.

 

[freeagent]

I am using Thermalright TF7 right now, it came with my cooler. Works pretty good so far..

 

[Nater]

Yeah I don't know what the deal is with those resellers, but the 30g tubes are sold through AliExpress (GD900 listings). It's so cheap there that they're practically giving it away. Right now they're US$2.60 each. Last time I ordered a couple in March it took three weeks to get them and the new batch performs just as good as the first one I got, so it seems like it performs consistently.



Yeah when it's 8 degrees hotter then it's not worth messing with. Pretty much every other thermal paste on the market does better than that stuff.

I'm thinking of doing some testing with my R9 270X to get some more accurate readings on these thermal pastes. It generates a good amount of heat and you can easily run the fans at 100% on a graphics card, so it might be a good candidate for it.

I'd still avoid it. Not one result for reviews from a reputable source in the first page of a google search on it. I'm not putting 10 cent thermal paste on my $450 CPU & $100 cooler. Hopefully someone out there takes the time to review it properly though, with it's value side. High TDP CPU, low tension/high tension brackets, waterblock vs air. etc. Then longevity. Does it "dry out"? I bet it does at that price.

 

[8BitZ80]

I'd still avoid it. Not one result for reviews from a reputable source in the first page of a google search on it. I'm not putting 10 cent thermal paste on my $450 CPU & $100 cooler. Hopefully someone out there takes the time to review it properly though, with it's value side. High TDP CPU, low tension/high tension brackets, waterblock vs air. etc. Then longevity. Does it "dry out"? I bet it does at that price.

Don't knock it 'till you've tried it. Obviously longevity takes a long time to thoroughly test but I've had it on my Ryzen 5600X since November 2020 (five months) and I've been using it for well over a year on all of my systems. Haven't had any issues with bad thermals or oil separation yet so I'm still waiting to find out how long each application can last. But already it's lasted a lot longer than I originally expected.

 

[nguyen]

Well the guy running TechYesCity is quite fond of the GD900 and called it the best paste for general usage, so it's not an untested paste.

 

[Shrek]

I use SuperLube thermal grease, it loses a few degrees but should last as it is good to 260C (I don't want to have to do the work more than once). This stuff is only $7 for 85g.

Concerning drying out; while I don't like thermal grease that dies out, that does no mean it is not still effective.

 

[8BitZ80]

Well the guy running TechYesCity is quite fond of the GD900 and called it the best paste for general usage, so it's not an untested paste.

Yeah correct, I actually found out about GD900 originally from his video and I ordered it in August 2019. About 1.5 years so that seems about right.

His results are similar to mine, in the sense that NT-H1 and GD900 scored the same results. I'd like to test more brands but that's all I have on hand at the moment.

 

[Shrek]

I've always wondered about diaper cream as that has zinc oxide in it, 40% in the maximum strength versions.

 

[8BitZ80]

I finally managed to find a slight difference between the two thermal pastes. Preliminary testing on the R9 270X resulted in a max of 49C for the GD900 and 50C for the NT-H1. Kind of an unofficial result though because the R9 270X doesn't produce as much heat as I wanted it to. I need to work on sorting out a better testbench and ordering some additional thermal paste options to test.

I've always wondered about diaper cream as that has zinc oxide in it, 40% in the maximum strength versions.

Eh, you can always try it on a system you don't care about. I tend to avoid wildly different products though because there can be issues with heat degradation or material incompatibilities.

 

[Shrek]

It was just an 'what if' moment... I'm happy with a silicone based zinc oxide mix; but liquid metal is something I will be avoiding after seeing what it does to aluminum.

 

[8BitZ80]

I did one more thorough test before I shelve this project. These results are very reliable, but AGAIN it's pretty much a draw.

Test regime:

1. Each thermal paste (NT-H1 and GD900) was applied on my Asus R9 270X DirectCUII graphics card
2. Next was an initial 10 minute burn-in with MSI Kombustor.
3. The main test was a 30 minute run using Unigine Heaven at 1080p Ultra
4. Ambient temperature was recorded at start and finish of the main run using an Elitech digital thermometer
5. A 5 minute cooldown with no load was used so that minimum temperatures could also be recorded
6. All GPU thermal results were recorded with HWiNFO64

Fan speeds on the card were set at 75% and I also used external fans to minimise heatsoak.

Results for NT-H1:

• Start ambient temp = 25.6C
• Finish ambient temp = 25.3C
• Maximum die temp = 51C
• Minimum die temp = 28C

Results for GD900:

• Start ambient temp = 25.7C
• Finish ambient temp = 25.9C
• Maximum die temp = 52C
• Minimum die temp = 29C

Spoiler: R9 270X thermal benchmark results

Noctua NT-H1:

GD900:

So the NT-H1 posted temperatures that were 1 degree cooler across the board, but it was also running in slightly cooler ambient temperatures (+0.4C). I can't confirm if the NT-H1 won by a full -1.0C margin because I can't get sub-decimal temperature readings on the GPU. There MIGHT be a -0.5C advantage to the NT-H1 but it's just as likely to be a dead equal result. Definitely there's no meaningful difference in the performance between the two of them.

 

[Ruru]

Getting probably some liquid metal today as I'll be picking up a box of stuff from my friend. Dunno if I have any usage to it myself.

There's also a delid tool on that box so my other friend's i7-6700K will be running cooler the next time when I visit him.

 

[VeqIR]

Where does one buy a tube of genuine Thermalright TFX? In the US or shipped to it.

 

[Liquid Cool]

I'll probably just continue to use Arctic Cooling's MX-2 as long as they keep selling it. I've been using the stuff for as long as I can remember and I really don't see the need to change.

It's always treated me well.

Best,

Liquid Cool

 

[Falkentyne]

Where does one buy a tube of genuine Thermalright TFX? In the US or shipped to it.

Nan's Gaming Gear and Corn's electronics both sell TFX in USA. (I've seen it on Amazon and Newegg). FrozenCPU can get it (Thermalright said FrozenCPU is one of their partners when I asked them in email).

TFX is extremely expensive however. The new kid on the block is SYY-157, which is less than half the price of TFX, performs exactly the same, and is also easier to spread while being the same overall composition as TFX, besides the viscosity and wetting improvements. FuzeIce Plus seems to be the exact same paste as SYY-157 in a side by side test and finger/spread test. I believe Maxtor CTG9 is also the exact same paste as well (CTG8 is worse). but is only available on aliexpress and is way too expensive (I looked carefully at the spread test of CTG9 and it behaves identically to SYY-157 in spatula spreading and adhesion and behavior, as well as color, so it's very likely the same as FuzeIce Plus/SYY-157. Both should come from the same or similar factories in Shenzhen, China.

But at $11 for 8 grams of SYY-157? That's a very good bargain. So I suggest trying SYY-157.

Amazon.com: Thermal Paste, SYY 8 Grams CPU Paste Thermal Compound Paste Heatsink for IC/Processor/CPU/All Coolers, 15.7W/m.k Carbon Based High Performance, Thermal Interface Material, CPU Thermal Paste : Electronics

Amazon.com: Thermal Paste, SYY 8 Grams CPU Paste Thermal Compound Paste Heatsink for IC/Processor/CPU/All Coolers, 15.7W/m.k Carbon Based High Performance, Thermal Interface Material, CPU Thermal Paste : Electronics

www.amazon.com

 

[VeqIR]

Some reviews I’ve seen put Noctua NT-H2 on par with Kingpin KPx and Thermalright TFX. Some variance in TG Kryonaut results too with respect to other top pastes. I have NT-H2 now (and a 1g tube of Kryonaut), so I’m wondering about KPx and TFX/SYY-157. P.S.: these don’t break my bank, and it’s fun, so I don’t mind getting other TIMs to compare.

 

[Borc]

For UK (or EU) users SYY-157 4g for just £2.60: https://www.amazon.co.uk/Grease，8-8...hild=1&keywords=syy-157&qid=1623413473&sr=8-5

 

[oldwalltree]

Some reviews I’ve seen put Noctua NT-H2 on par with Kingpin KPx and Thermalright TFX. Some variance in TG Kryonaut results too with respect to other top pastes. I have NT-H2 now (and a 1g tube of Kryonaut), so I’m wondering about KPx and TFX/SYY-157. P.S.: these don’t break my bank, and it’s fun, so I don’t mind getting other TIMs to compare.

I had NT-H2 on my 10900x and switched to XPX and well.....there was no difference!

 

[r.h.p]

im not sure what this stuff is but it works good and is cheap from the electronics store

 

[PaulieG]

I'm wondering if I should buy 3-4 of the lesser known pastes and do a comparison review of them against my "go to" paste NT-H2 and maybe one other like TFX, all on one of my 5950x's. That would be a fun late fall/winter project when I'm not outside as much.