What Paste Application Method do You Use in 2020?

What Paste Application Method do You Use in 2020?

  • Pea sized ball in center

    Votes: 78 36.3%
  • X accross heatspreader or die

    Votes: 34 15.8%
  • Spread evenly (credit card, or other straight edge)

    Votes: 62 28.8%
  • Spread evenly (using finger wearing gloves)

    Votes: 15 7.0%
  • other

    Votes: 12 5.6%
  • Grain of rice

    Votes: 14 6.5%

  • Total voters
    215
I've heard with the new higher end Ryzen 3000 chips that an X, full spread or H shape is more optimal. Based on what you all have done, have you seen a difference with either spread style?
I'm seeing higher temps on my 3900x on an ASUS Strix x570 board and have already done with undervolting by .100 to drop temps some. Next up is likely a remount, but just curious on your thoughts. Thanks
 
I've always done a full spread with a plastic card, particularly since I used to always use Arctic Silver which is incredibly thick paste. I've got Arctic MX-4 now which isn't as thick, but still prefer to spread it myself to eliminate any possibility of uneven coating.
 
Thermal Grizz Liquid Metal spread over the area the cpu die is on both the IHS and Water Block (also on the die itself if delidded). Once you go liquid metal you never go back, the risk is overblown, it's not runny or anything once spread, just don't apply on any leads or other components like a dummy to short your shit. The gainz are worfz it. :p
 
I use credit card to put thin layer paste and after that I put a little more paste in the center.
 
Last edited:
I chose the pea dropped because I have used that for decades, but the larger chips, you might use two p drops...
 
Going to try Thermal Grizzly Carbonaut Thermal Pad first, if temperatures are below throttling decently, call it a day. Tempted to lap the CPU heat spreader and block, probably wait to see if needed, CPU is good etc.
https://www.amazon.com/Thermal-Griz...Thermal+Pad+51x68x0.2mm&qid=1581984375&sr=8-2
That is expensive. It's hard for me to justify the cost to even experiment, when I have 2 tubes of thermal paste (MX-4 and PK-3) that, together, cost less than the one pad.

What were your results?

This is a really good resource!

I've always done the pea-sized blob in the middle. It's always served me well; I tried to spread paste in my most recent re-application of paste and found it to be a ton of work. So, I stopped in the middle and just went back to pea-sized. Looking at the above, the difference is negligible, so the amount of labor required just didn't seem worth it in the end.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
Spread on thin, wipe off, then (small) pea sized dot in center. Initial spread fills in the air gaps you can't see, and then the dot in the center spreads with pressure to fill in the rest.
 
That is expensive. It's hard for me to justify the cost to even experiment, when I have 2 tubes of thermal paste (MX-4 and PK-3) that, together, cost less than the one pad.

What were your results?

This is a really good resource!

I've always done the pea-sized blob in the middle. It's always served me well; I tried to spread paste in my most recent re-application of paste and found it to be a ton of work. So, I stopped in the middle and just went back to pea-sized. Looking at the above, the difference is negligible, so the amount of labor required just didn't seem worth it in the end.
BeQuiet Dark rock 4, 3960x, MSI TRX40 Pro WiFi, room temp avg 20c, Aida64 stress test with avx, believe over 9 hours with avg of 92c. CPU never thermal throttled, at least HWInfo indicated it didn't, avg CPU clock speed 3.91 hz which is above base clock. 280w. Test 100% successful as well as no WHET errors. That is using the Carbonaut thermal pad.

While usable, losing boost frequency due to higher temperature., losing performance. Not interested in trying paste since once board is put into case it will use a custom loop. Paste could give a 2c-4c advantage but for me it is still too warm.
 
I always used multiple smaller drops, I found one bigger blob in the middle results in the paste not spreading properly, and ends up as a thick layer in the middle while the corners remain dry.
I also turn and twist and move the sink on the cpu back and forward a bit to make the paste spread evenly. Worked so far.
Only time I used the one drop to the middle was for the old naked die cpus, like socket A and socket 370.
 
The best seating resulting in the best temps and performance I got out of my 3900x was a blob in the middle. Here's a link to another one of my posts showing what it looked like when I removed the cooler. 3900x thermal compound and seating
That's a damned near-perfect blob result! My most recent switch from MX-4 to Pk3 showed quite a lot of dry area around the corners from my MX-4 application. Clearly I didn't use enough paste the first time, but "rice-sized" is a bit smaller than "pea-sized" and, even then, "pea-sized" is a hard measurement for me to get right. Do I use a small pea, large pea, green pea, snowpod pea, etc as my size measurement? :p

I don't know how I missed this post the first time reading through this thread. I may have to go with the "X" or double-line method on my next thermal paste application, looking at this video. I have an i7-3930k and have always referred to Arctic's instructions, alongside general comments, to determine the center blob as my method. However, as stated above, my last MX-4 application looked like the first blob, but smaller with quite some dry area at the edges.
 
That's a damned near-perfect blob result! My most recent switch from MX-4 to Pk3 showed quite a lot of dry area around the corners from my MX-4 application. Clearly I didn't use enough paste the first time, but "rice-sized" is a bit smaller than "pea-sized" and, even then, "pea-sized" is a hard measurement for me to get right. Do I use a small pea, large pea, green pea, snowpod pea, etc as my size measurement? :p

It was what I would call a small sized pea I guess ;)

On the waterblock that is attached now, I did a thin layer with a credit card.
 
I plan on building my new Ryzen 5800X rig next week so was doing some research on 'correct' thermal paste application...the X method seems best for larger Threadripper type of CPU's while the pea sized drop in the middle seems like the safest bet for all others...either way the most important thing is to not apply too much
 
I do the dot in the center and use a credit card to spread it out evenly. Then I plop the block down, remove and check the footprint. Usually I get a perfect coverage.
 
I'm surprised there aren't plastic/rubber accessories like dental dams fit to processors; Lay down a rubber shield, apply slightly too much such that proper pressure will cause a tiny excess to squeeze out, but pull away the rubber like masking tape and you have edge to edge coverage as well as no spillage.
 
  • Like
Reactions: x509
like this
Dot or X. If you use too much paste, a cooler with proper pressure will squeeze any excess paste out.
 
I ordered a cpu thermal pad to try out. It seems I swap cpus or heatsinks often enough that between the difficulty to remove the heatsink without ripping out am4 cpus and the time it takes to clean everything once I do, I don't mind sacrificing a couple degrees when compared to a perfect application, which who knows if I ever really got with thermal paste.
 
difficulty to remove the heatsink without ripping out am4 cpus
youre doin it wrong.
boot to bios for 5 minutes to warm up chip/paste, power off then twist the heatsink. no stuck cpu. been that way since they put an ihs on em. its even in the instructions....
 
youre doin it wrong.
boot to bios for 5 minutes to warm up chip/paste, power off then twist the heatsink. no stuck cpu. been that way since they put an ihs on em. its even in the instructions....
I'm aware, trust me. It might come down to the paste, I use noctua nt h2. I always do it warm (system usually on) and I rotate the cooler as much as possible. Not all heatsinks can be rotated more than a few degrees with vrm heatsinks or ram clearances. The noctua paste is fairly sticky in this case.
 
I'm aware, trust me. It might come down to the paste, I use noctua nt h2. I always do it warm (system usually on) and I rotate the cooler as much as possible. Not all heatsinks can be rotated more than a few degrees with vrm heatsinks or ram clearances. The noctua paste is fairly sticky in this case.
using a pad here ever since my cpu pulled ou the socket and ALMOST got that crap on my pins (can you imagine that crap getting in the socket? :eek: ).....Your right To much is a TERRIBLE disaster waiting to happen. The pads never dry out and have zero risk and as far as i can tell get good results....not liquide metal or bare die results but not bad at all.
 
I follow the Arctic Silver method. I learned a lot reading their documentation.

I also use and recommend their cleaner to remove old goop. This stuff is amazing.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
I heavily microwave a mug full of water, and once it's BOILING hot, I drop in my syringe of thermal paste.
I let it get nice and hot. Very hot.
Then I use a heatgun or hot blow-dryer on the heatsink, and even give a couple passes on the CPU itself.

Once the metal is nice and HOT, and the paste is hot, I gloop on ONE GIANT DROP. Like a pea to end all peas.
Then I push the heatsink straight down, perfectly flat, and sploooooosh~!

Perfection.
 
  • Like
Reactions: Nobu
like this
X marks the spot, since Nehalem 1366 days.

Remember Socket 462 Athlons? And FCPGA s-370s? Those were fun. ;-) Saw lots of carnage pics, never chipped or killed one personally.
 
I think for most pastes and CPU's I use the pea sized ball method and mash it out with the heatsink.

Thermal Grizzly Kryonaut has been a little thicker though, and I don't think that is a good method for that paste. Also the Threadripper is large, also making this method not suitable.

Last time I tried spreading it thin while wearing a plastic glove. Seems to have worked well!
 
I'm aware, trust me. It might come down to the paste, I use noctua nt h2. I always do it warm (system usually on) and I rotate the cooler as much as possible. Not all heatsinks can be rotated more than a few degrees with vrm heatsinks or ram clearances. The noctua paste is fairly sticky in this case.

I pulled a 3900x right out of the socket (thankfully OK) with D15 and NTH1 i had both warmed it up and twisted prior to trying to remove the cooler. Not going to use noctua anymore due to its stickiness
 
I pulled a 3900x right out of the socket (thankfully OK) with D15 and NTH1 i had both warmed it up and twisted prior to trying to remove the cooler. Not going to use noctua anymore due to its stickiness
yeah, you can rotate it all around but it doesn't seem to break free. I moved on to a thermal pad since I seem to change my cooling so frequently, saves on materials and cleanup and eliminates the risk.
 
Historically, I've always used Pea or X. So long as the cooler applies sufficient pressure you'll end up with a sufficiently thin layer. It's much more detrimental to trap air than to end up with more paste than needed.

But with last build I followed Arctic's instructions for the cooler I was using. The gaps between the heatpipes would likely compromise the squeezeout pattern of a single dot or X, but also seem sufficient to prvide a venting path for any air that would otherwise get trapped by this application method:
1609127281667.png
 
For my latest application I used a hair dryer to warm up both the tube of Thermal Grizzly Kryonaut AND the surface of the installed CPU before starting (Kryonaut is difficult to spread when cold)

Then I pulled on a nitrile glove put some paste on the CPU and spread it as thin as I could with my finger.

Had to make sure I scraped my finger off on the side of the CPU, as lifting straight up caused th epaste to stick, and left a "bald spot" on the CPU.

Very happy with the result. Probably the best paste distribution I've ever had, and the Kryonaut is notorious for being difficult to spread.
 
Initially I started with the credit card spread with AS5, but then I moved to the rice/small pea size when I moved to MX-4. I was noticing I wasn't getting good coverage with the HSF application, so I started using the X shape. X works best for me with MX-4.
 
Initially I started with the credit card spread with AS5, but then I moved to the rice/small pea size when I moved to MX-4. I was noticing I wasn't getting good coverage with the HSF application, so I started using the X shape. X works best for me with MX-4.

X shape is indeed the best, and is absolutely essential with extremely thick pastes like Thermalright TFX (that's why Luumi didn't get good results with it). TFX is literally the best paste on the market, even though it starts off a C or two worse than Kryonaut, it cures as it turns a bit more 'putty like' after about a week, reducing temps even more, and it is as endurable as old school Ceramique. This makes it ideal for video cards without flat GPU surfaces like has been reported on Ampere. And unless you're using LM on a laptop, TFX should be the only paste worth considering.

https://forums.guru3d.com/threads/t...ivity-of-14-2-w-mk.435511/page-2#post-5861772
 
X shape is indeed the best, and is absolutely essential with extremely thick pastes like Thermalright TFX (that's why Luumi didn't get good results with it). TFX is literally the best paste on the market, even though it starts off a C or two worse than Kryonaut, it cures as it turns a bit more 'putty like' after about a week, reducing temps even more, and it is as endurable as old school Ceramique. This makes it ideal for video cards without flat GPU surfaces like has been reported on Ampere. And unless you're using LM on a laptop, TFX should be the only paste worth considering.

https://forums.guru3d.com/threads/t...ivity-of-14-2-w-mk.435511/page-2#post-5861772


I don't put much faith in the paste shootouts.

So much of the difference between pastes is really a difference from mount to mount that is often skewing the data.

In order to do it right, it would be a lot of work, and no reviewer in their right mind is going to do it.

I'm talking a statistically significant number of heatsink mounts with each paste. With no historical data to go on, I'd start with 30 mounts per paste, and then do statistical analysis on the data, comparing means, standard deviations, and computing K-factor tolerance intervals.

If someone did this, I'd pay attention, but until that point my feeling is that paste really doesn't matter much. All of the contenders are really close to eachother if you get a good mount. Heck, FrgMstr himself used to use the good old white goop (a bulk packaged version of the type of stuff we used to get in blister packs with heatsinks in the 90's) in most of his reviews.
 
I don't put much faith in the paste shootouts.

So much of the difference between pastes is really a difference from mount to mount that is often skewing the data.

In order to do it right, it would be a lot of work, and no reviewer in their right mind is going to do it.

I'm talking a statistically significant number of heatsink mounts with each paste. With no historical data to go on, I'd start with 30 mounts per paste, and then do statistical analysis on the data, comparing means, standard deviations, and computing K-factor tolerance intervals.

If someone did this, I'd pay attention, but until that point my feeling is that paste really doesn't matter much. All of the contenders are really close to eachother if you get a good mount. Heck, FrgMstr himself used to use the good old white goop (a bulk packaged version of the type of stuff we used to get in blister packs with heatsinks in the 90's) in most of his reviews.

I know but people's results don't lie.
If Kryonaut pumped out (temps rose) within 1 month on a laptop, and TFX was stable 6 months+, then that means TFX is better for that system. Temp comparisons are meaningless if you have to repaste after a month because temps start rising.
I noticed the same thing on my 3090 FE, with the new Kryonaut Extreme. Temps rising after a few weeks. While before I ran out of TFX, temps were always stable.

https://www.igorslab.de/en/geforce-...water-and-air-coolers-so-heavy-investigative/

So I think that has something to do with it.
Having a top performing thermal paste doesn't mean that much if it isn't stable without having to repaste often because of uneven pressure.
 
With Intel I just put a line down the center over the cores. I am on AMD now so still experimenting. But +1 for TFX, man this stuff is pretty decent. I was using TF8 previously, but ran dry. I am almost out of TFX now.. just got a 2G tube, but will order a big one probably tonight. It is pretty thick though.. experimenting is expensive lol.
 
I know but people's results don't lie.
If Kryonaut pumped out (temps rose) within 1 month on a laptop, and TFX was stable 6 months+, then that means TFX is better for that system. Temp comparisons are meaningless if you have to repaste after a month because temps start rising.
I noticed the same thing on my 3090 FE, with the new Kryonaut Extreme. Temps rising after a few weeks. While before I ran out of TFX, temps were always stable.

https://www.igorslab.de/en/geforce-...water-and-air-coolers-so-heavy-investigative/

So I think that has something to do with it.
Having a top performing thermal paste doesn't mean that much if it isn't stable without having to repaste often because of uneven pressure.

Huh. I had Kryonaut in my last build for 3 years without replacing it and never noticed any temperature increases.
 
  • Like
Reactions: x509
like this
Back
Top