[EOL] Arctic MX-5 is here!!Tests incoming! Completed. Now its MX-6 testing time!

[lexluthermiester]

Doesn't that happen with all compounds eventually?

Depends on the compound.

I thought drying/caking was bad. It happened to me with AS5.

Mine killed a gpu, but back then they only held it down with spring loaded pins.

If it was a video card that didn't have a fan or had a crap fan, that's what caused the card to die. AS5 drying out had nothing to do with it. AS5 stays viable and function even when dried out. AS5 generally dries out 6 to 8 months after being installed. This is by design from what I understand.

 

[mtcn77]

AS5 generally dries out 6 to 8 months after being installed. This is by design from what I understand.

That must be some design that eludes me.
I still carry it around, though. Computers are always in short supply around me.

 

[Space Lynx]

even though I am getting 2 celsius worse, I suppose that is one benefit of mx5 paste... i really don't have to worry about changing this paste for like 10 yrs LOL

 

[Shrek]

My batch of GD900 finally got here; so I could do some tests on a video card

• Dry 84°C
• SuperLube thermal compound 72°C
• Petroleum Jelly 71°C
• Diaper cream (40%) 68°C
• GD900 68°C

Perhaps this will convince someone to include Petroleum Jelly and Diaper cream in their tests,
just for comparison as the Diaper cream would soon dry out.

Things I didn't try

• Dielectric grease
• Anti-seize grease

 

[lexluthermiester]

Perhaps this will convince someone to include Petroleum Jelly and Diaper cream in their tests

Nope. But I'm intrigued by your results. Did you do screen caps?

 

[Shrek]

This was just idle as FurMark no longer runs on this card

 

[Mussels]

This was just idle as FurMark no longer runs on this card

So its idle tests with no ambient results, no fan speeds, no other relevant information?


Your tests may be of interest to you, but they're so lacking in detail they're not of much functional use

 

[Shrek]

They were intended to see if Petroleum Jelly and Diaper cream could do the job, that is all;
with GD900 for comparison.

I found this useful in that it told me Diaper cream was as good as GD900 and so perhaps
longevity is a major factor when comparing thermal greases and not just that one last
degree cooler.

 

[8BitZ80]

Time to settle the score and bust some myths. I tested MX-4, MX-5, NT-H1, GD900 and KPX on my delidded overclocked Core i5 4670K running at 4.4GHz @1.35V with a Hyper 212X cooler and fans on 100%. I ran the Furmark CPU burner for 2 minute runs with 1 minute breaks in between to get repeatable temperature readings. Each TIM was applied on both the die and the IHS to magnify the slight differences in performance as much as possible. Everything was cleaned with IPA as best I could and I made it as level a playing field as humanly possible. Wall power draw for each run was hovering around 170 watts (I was using the iGPU).

I used HWInfo64 for data logging and these were my maximum temperatures for all four cores on each run:

Arctic MX-4 Run 1 (Ambient 23.1C) = 77C, 76C, 76C, 74C
Arctic MX-4 Run 2 (Ambient 23.2C) = 78C, 77C, 76C, 74C
Arctic MX-4 Run 3 (Ambient 23.2C) = 78C, 77C, 76C, 74C

Arctic MX-5 Run 1 (Ambient 23.6C) = 72C, 71C, 71C, 70C
Arctic MX-5 Run 2 (Ambient 23.2C) = 73C, 73C, 73C, 72C
Arctic MX-5 Run 3 (Ambient 23.3C) = 73C, 72C, 73C, 72C

Noctua NT-H1 Run 1 (Ambient 23.6C) = 70C, 70C, 69C, 68C
Noctua NT-H1 Run 2 (Ambient 23.2C) = 71C, 72C, 72C, 70C
Noctua NT-H1 Run 3 (Ambient 23.3C) = 71C, 72C, 72C, 70C

GD900 Run 1 (Ambient 23.2C) = 72C, 72C, 72C, 71C
GD900 Run 2 (Ambient 22.9C) = 75C, 74C, 76C, 73C
GD900 Run 3 (Ambient 23.3C) = 76C, 75C, 75C, 73C

Kingpin KPX Run 1 (Ambient 23.1C) = 71C, 71C, 72C, 70C
Kingpin KPX Run 2 (Ambient 23.3C) = 70C, 69C, 70C, 68C
Kingpin KPX Run 3 (Ambient 22.9C) = 70C, 70C, 71C, 70C

Keep in mind that the first run is essentially a warm-up. Run 2 and Run 3 results are more repeatable and more indicative of the actual performance.

 

[Shrek]

Looks like MX-5 has found its match...

Looking forward to the GD900 runs (no GD900-1?) as I now have enough for a lifetime.

 

[8BitZ80]

Looks like MX-5 has found its match...

Looking forward to the GD900 runs (no GD900-1?) as I now have enough for a lifetime.

Results are updated. Yeah I tried GD900-1 previously and it was nowhere near as good as regular GD900, so it wasn't included. GD900 put on a good show though and outperformed MX-4.

 

[Space Lynx]

Time to settle the score and bust some myths. I tested MX-4, MX-5 and NT-H1 on my delidded overclocked Core i5 4670K running at 4.4GHz @1.35V with a Hyper 212X cooler and fans on 100%. I ran the Furmark CPU burner for 2 minute runs with 1 minute breaks in between to get repeatable temperature readings. Each TIM was applied on both the die and the IHS to magnify the slight differences in performance as much as possible. Everything was cleaned with IPA as best I could and I made it as level a playing field as humanly possible. Wall power draw for each run was hovering around 170 watts (I was using the iGPU).

I used HWInfo64 for data logging and these were my maximum temperatures for all four cores on each run:

Arctic MX-4 Run 1 (Ambient 23.1C) = 77C, 76C, 76C, 74C
Arctic MX-4 Run 2 (Ambient 23.2C) = 78C, 77C, 76C, 74C
Arctic MX-4 Run 3 (Ambient 23.2C) = 78C, 77C, 76C, 74C

Arctic MX-5 Run 1 (Ambient 23.6C) = 72C, 71C, 71C, 70C
Arctic MX-5 Run 2 (Ambient 23.2C) = 73C, 73C, 73C, 72C
Arctic MX-5 Run 3 (Ambient 23.3C) = 73C, 72C, 73C, 72C

Noctua NT-H1 Run 1 (Ambient 23.6C) = 70C, 70C, 69C, 68C
Noctua NT-H1 Run 2 (Ambient 23.2C) = 71C, 72C, 72C, 70C
Noctua NT-H1 Run 3 (Ambient 23.3C) = 71C, 72C, 72C, 70C

GD900 Run 1 (Ambient 23.2C) = 72C, 72C, 72C, 71C
GD900 Run 2 (Ambient 22.9C) = 75C, 74C, 76C, 73C
GD900 Run 3 (Ambient 23.3C) = 76C, 75C, 75C, 73C

Kingpin KPX Run 1 (Ambient 23.1C) = 71C, 71C, 72C, 70C
Kingpin KPX Run 2 (Ambient 23.3C) = 70C, 69C, 70C, 68C
Kingpin KPX Run 3 (Ambient 22.9C) = 70C, 70C, 71C, 70C

Keep in mind that the first run is essentially a warm-up. Run 2 and Run 3 results are more repeatable and more indicative of the actual performance.

I really wonder if you got a bad batch or one of those fake mx-4 pastes... that is a lot of discrepancy in temps... especially considering i do better with mx-4 over mx-5.

 

[nguyen]

I really wonder if you got a bad batch or one of those fake mx-4 pastes... that is a lot of discrepancy in temps... especially considering i do better with mx-4 over mx-5.

Did you use GPU to test the MX5? might need some more pressure on it then, do some gaming then tighten the mounting screws just a little bit more when the GPU is still warm.
Higher viscosity paste just need higher mounting pressure to squeeze out the unnecessary part since smaller gap offer better thermal transfer.

 

[8BitZ80]

I really wonder if you got a bad batch or one of those fake mx-4 pastes... that is a lot of discrepancy in temps... especially considering i do better with mx-4 over mx-5.

There will be some big differences, but that's the whole point of applying each TIM on both the die and IHS. In reality with a single application the differences would be within margin of error and that makes it very difficult to compare them. A bad-performing thermal paste might be up around 85-90C with this setup.

With a fair amount of certainty I can say that none of the pastes are fake. I bought the Arctic pastes from one of the most trusted retailers in the country (PC Case Gear Australia). The GD900 is also not fake (I know because I've bought and tested multiple batches) and MX-4 performs similar to GD900, which makes sense based on what I've seen from other reviews.

 

[Space Lynx]

Did you use GPU to test the MX5? might need some more pressure on it then, do some gaming then tighten the mounting screws just a little bit more when the GPU is still warm.
Higher viscosity paste just need higher mounting pressure to squeeze out the unnecessary part since smaller gap offer better thermal transfer.

no I own the rx 6800 non-xt, you don't replace the paste on that. i have never opened mine up, its carbon based and if you open it up you ruin it.

i am only talking cpu's.

 

[AOne]

And proper results would show up after 2-3 weeks of hard use. Older pastes have much lower maximum working temps and would perform good in the first few days and then rapidly (!!) degrade. Newer pastes like Noctua NT-H2, Kingpin KPX, etc, have a max working temp of 200 C and this issue in not affecting them. In general - if your testing results are not repeated 2 weeks later that's only half the story. Been through this on few laptops with MX4, AS5, LM, some unknown stuff found at home and so on. If your chips are working hot, like on all the recent years laptops, it's a really noticeable problem.

 

[8BitZ80]

And proper results would show up after 2-3 weeks of hard use. Older pastes have much lower maximum working temps and would perform good in the first few days and then rapidly (!!) degrade. Newer pastes like Noctua NT-H2, Kingpin KPX, etc, have a max working temp of 200 C and this issue in not affecting them. In general - if your testing results are not repeated 2 weeks later that's only half the story. Been through this on few laptops with MX4, AS5, LM, some unknown stuff found at home and so on. If your chips are working hot, like on all the recent years laptops, it's a really noticeable problem.

At least with the pastes that I've used on my main rig (MX-4, MX-5 and GD900) I haven't detected any changes in performance over time. I normally find that testing weeks-old applications and then re-applying results in practically no change with standard pastes. That said, I generally try to avoid benchmarking performance changes over time because it's impossible to control the scope. There's far too many variables and inevitably everyone will find a bone to pick when it comes to the testing methodology.

 

[Space Lynx]

im honestly ok with my mx-5, just disappointed is all. but yeah, its kind of a pain to change the paste, so don't think I will do so again, just accept my extra 2 deg i gained... not end of world. and like I said before, mx-5 has added benefit of being fresh off the production line so I know for a fact it will last 8 years without repasting, and honestly where I am with this hobby. that's probably a good thing. my upgrading days are done I think... more to life than games.

 

[Shrek]

There will be some big differences, but that's the whole point of applying each TIM on both the die and IHS. In reality with a single application the differences would be within margin of error and that makes it very difficult to compare them. A bad-performing thermal paste might be up around 85-90C with this setup.

But doesn't that just mean you have applied them too thick and given advantage to the runny pastes that are not good against pump out?

Results are updated. Yeah I tried GD900-1 previously and it was nowhere near as good as regular GD900, so it wasn't included. GD900 put on a good show though and outperformed MX-4.

That is incredibly useful information, given how expensive GD900-1 is compared to GD900

 

[nguyen]

But doesn't that just mean you have applied them too thick and given advantage to the runny pastes that are not good against pump out?

Viscosity doesn't really matter with CPU cooler with high mounting pressure, the problem only exist with GPU and Laptop cooler where mounting pressure is Medium to Low.

 

[Shrek]

But PooPipeBoy said

" ... that's the whole point of applying each TIM on both the die and IHS. In reality with a single application the differences would be within margin of error and that makes it very difficult to compare them."

i.e. thicker gave differing results.

 

[nguyen]

But PooPipeBoy said

" ... that's the whole point of applying each TIM on both the die and IHS. In reality with a single application the differences would be within margin of error and that makes it very difficult to compare them."

i.e. thicker gave differing results.

What he meant was had he changed only the TIM on top of the IHS, the difference between all the tested TIM would be within margin of errors (1-2C).
High performance TIM only show tangible benefits when applied to bare die (as with GPU and Laptop), for CPU with IHS there are not much difference between the worst and the best TIM.

 

[8BitZ80]

But PooPipeBoy said

" ... that's the whole point of applying each TIM on both the die and IHS. In reality with a single application the differences would be within margin of error and that makes it very difficult to compare them."

i.e. thicker gave differing results.

I haven't seen any results that would support that theory. In fact the Kingpin KPX paste was actually the most viscous and MX-4/GD900 are very easy to spread.

Viscosity doesn't really matter with CPU cooler with high mounting pressure, the problem only exist with GPU and Laptop cooler where mounting pressure is Medium to Low.

That seems to be correct. When I've used my R9 270X it actually helps MX-4 to beat Kingpin KPX, which I think just comes down to the lower mounting pressure changing the pecking order. The type of testbenches and heatsinks used can make a big difference in that regard.

 

[r.h.p]

Depends on the compound.


If it was a video card that didn't have a fan or had a crap fan, that's what caused the card to die. AS5 drying out had nothing to do with it. AS5 stays viable and function even when dried out. AS5 generally dries out 6 to 8 months after being installed. This is by design from what I understand.

dont have drying problems with the electronics store paste

 

[lexluthermiester]

no I own the rx 6800 non-xt, you don't replace the paste on that. i have never opened mine up, its carbon based and if you open it up you ruin it.

I did not know this. Very interesting..

im honestly ok with my mx-5, just disappointed is all. but yeah, its kind of a pain to change the paste, so don't think I will do so again, just accept my extra 2 deg i gained... not end of world. and like I said before, mx-5 has added benefit of being fresh off the production line so I know for a fact it will last 8 years without repasting, and honestly where I am with this hobby. that's probably a good thing.

Maybe the issue you're having it specific to your setup. How do you apply the TIM? Are you using the system in your specs? Is there anything special you might be doing that may be causing a difference?

my upgrading days are done I think... more to life than games.

While true, life is so much less fun without them..

which I think just comes down to the lower mounting pressure changing the pecking order.

This. Mounting pressure matters and can have a measurable effect, even an effect larger than one might expect.