• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Arctic MX-5 is here boys!!!!! Tests incoming!!!

Joined
Apr 9, 2018
Messages
771 (0.66/day)
I did a shootout between MX-2, MX-4 and MX-5. Results are very interesting.

This is no ordinary test because I used each paste on the die AND the IHS of a delidded processor. That means that the heat needs to transfer through two applications of the same paste and it really helps to highlight the tiny differences in performance.

Testbench is a delidded 120W TDP Xeon X5460 at 3.6GHz 1.35V with a 6 heatpipe Snowman 92mm dual tower cooler. Fan speeds are at maximum (molex powered). I used HWInfo64 to log data across two-minute runs of the Furmark CPU burner with 1 minute breaks in between. Ambients changed only slightly in the process but I recorded it during each run.

MX-2 Run 1 (24.3C Ambient) = 70C, 64C, 71C, 70C
MX-2 Run 2 (24.3C Ambient) = 70C, 64C, 70C, 69C
MX-2 Run 3 (24.3C Ambient) = 70C, 64C, 70C, 69C

MX-4 Run 1 (24.5C Ambient) = 73C, 67C, 73C, 72C
MX-4 Run 2 (24.5C Ambient) = 72C, 67C, 73C, 72C
MX-4 Run 3 (24.5C Ambient) = 73C, 67C, 73C, 72C

MX-5 Run 1 (24.7C Ambient) = 71C, 66C, 73C, 70C
MX-5 Run 2 (24.7C Ambient) = 70C, 64C, 72C, 69C
MX-5 Run 3 (24.8C Ambient) = 70C, 65C, 72C, 69C

I'll keep conclusions to a minimum but there definitely is a performance improvement with MX-5. Impressive numbers from MX-2 as well.

I also thought that all three pastes were very easy to spread. Wouldn't hesitate using all of them. MX-5 is very light blue in color and is hard to see on paper towel, but that's only a minor downside.
 
Joined
Mar 21, 2021
Messages
465 (5.54/day)
System Name HP Compaq 8000 Elite CMT
Processor Intel Core 2 Quad Q9550
Motherboard Hewlett-Packard 3647h
Memory 16GB DDR3
Video Card(s) NVIDIA GeForce GT 1030 GDDR5 (fan-less)
Storage 2TB Seagate Firecuda 3.5"
Display(s) Dell P2416D (2560 x 1440)
Power Supply 12V HP proprietary
Software Windows 10 Pro 64-bit
Joined
Oct 17, 2014
Messages
6,931 (2.85/day)
Location
USA
Processor i7-7820HK
Memory 2x8gb (16gb)
Video Card(s) gtx 1070
Storage 2TB SSD
Display(s) 17.3" 96hz 1080p IPS
I did a shootout between MX-2, MX-4 and MX-5. Results are very interesting.

This is no ordinary test because I used each paste on the die AND the IHS of a delidded processor. That means that the heat needs to transfer through two applications of the same paste and it really helps to highlight the tiny differences in performance.

Testbench is a delidded 120W TDP Xeon X5460 at 3.6GHz 1.35V with a 6 heatpipe Snowman 92mm dual tower cooler. Fan speeds are at maximum (molex powered). I used HWInfo64 to log data across two-minute runs of the Furmark CPU burner with 1 minute breaks in between. Ambients changed only slightly in the process but I recorded it during each run.

MX-2 Run 1 (24.3C Ambient) = 70C, 64C, 71C, 70C
MX-2 Run 2 (24.3C Ambient) = 70C, 64C, 70C, 69C
MX-2 Run 3 (24.3C Ambient) = 70C, 64C, 70C, 69C

MX-4 Run 1 (24.5C Ambient) = 73C, 67C, 73C, 72C
MX-4 Run 2 (24.5C Ambient) = 72C, 67C, 73C, 72C
MX-4 Run 3 (24.5C Ambient) = 73C, 67C, 73C, 72C

MX-5 Run 1 (24.7C Ambient) = 71C, 66C, 73C, 70C
MX-5 Run 2 (24.7C Ambient) = 70C, 64C, 72C, 69C
MX-5 Run 3 (24.8C Ambient) = 70C, 65C, 72C, 69C

I'll keep conclusions to a minimum but there definitely is a performance improvement with MX-5. Impressive numbers from MX-2 as well.

I also thought that all three pastes were very easy to spread. Wouldn't hesitate using all of them. MX-5 is very light blue in color and is hard to see on paper towel, but that's only a minor downside.

I used to have some mx-2... lost it a long time ago. looks like mx-2 beats mx-5. lol...
 
Joined
Apr 9, 2018
Messages
771 (0.66/day)
If you have a heat-gun could you bake them to give some sense of how they might hold up with age?

I'm not sure how to test for the so called 'pump out effect'
Der8auer Deep-Dives on Thermal Paste: Misconceptions, Curing, & More | LTX 2019 - Bing video

From what I've seen there doesn't seem to be enough reason to test longevity and it would be very difficult to make a fair test. I'm thinking of testing the curing effect of Arctic Silver 5, but only because it's already well known that curing is necessary to get good performance with ceramic-based pastes. Like Der8auer said it seems like most people confuse pump-out and curing.

I used to have some mx-2... lost it a long time ago. looks like mx-2 beats mx-5. lol...

On a couple of cores the MX-2 schooled the newer pastes, yeah. I can't remember which tests but I have seen other reviewers post great numbers with MX-2 so I was kind of expecting it to put on a good show.
 
Joined
Jan 21, 2019
Messages
54 (0.06/day)
System Name CS Prime
Processor i7-5775C @4ghz eDRAM @ 2ghz
Motherboard Z97 Sabertooth USB 3.1 Mk1
Cooling Corsair H90
Memory Kingston HyperX Fury 32GB DDR3L @ 2ghz x2
Video Card(s) Gigabyte Windforce GTX1080
Storage 512GB NVMe MyDigitalSSD BPXP, 4TB SATA WD Blue
Display(s) Q900RB
Audio Device(s) Denon AVR-X3700H
Power Supply Seasonic PRIME 650
Mouse MX Master
Keyboard MX Keys
Joined
Oct 17, 2014
Messages
6,931 (2.85/day)
Location
USA
Processor i7-7820HK
Memory 2x8gb (16gb)
Video Card(s) gtx 1070
Storage 2TB SSD
Display(s) 17.3" 96hz 1080p IPS
On a couple of cores the MX-2 schooled the newer pastes, yeah. I can't remember which tests but I have seen other reviewers post great numbers with MX-2 so I was kind of expecting it to put on a good show.

Seems like a bad company that produces same product but worse 6 years after the best one... or whenever it came out. lol

oh well. at least i know its fresh and i got there in-case of emergency someday.
 
Joined
Jun 30, 2014
Messages
108 (0.04/day)
The usual fight for 1-2-3 degrees that make the difference :)
In price also
 
Joined
Jun 18, 2015
Messages
313 (0.14/day)
Location
Perth , West Australia
System Name schweinestalle
Processor AMD Ryzen 7 3700 X
Motherboard Asus Prime - Pro X 570 + Asus PCE-AC68 Dual Band PCI-E Wi-Fi Adapter
Cooling Standard Air
Memory Kingston HyperX 2 x 8 gb DDR 4 3200mhz
Video Card(s) AMD Radeon 5700 XT 8 GB Strix
Storage Intel SSD 240 gb Speed Demon & WD 240 SSD Blue & WD 250 SSD Green & Seagate Barracuda 1 TB Sata
Display(s) Asus XG 32 V ROG
Case Corsair AIR ATX
Audio Device(s) Realtech standard
Power Supply Corsair 850 Modular
Mouse CM Havoc
Keyboard Corsair Cherry Mechanical
Software Win 10
Benchmark Scores Unigine_Superposition 4K ultra 7582
im curious if anyone who actually should have their hands inside of a PC, has ever damaged something with Arctic Silver5. (rhetorical, im certain a million fools have damaged a million components all in equally ridiculous way)
ive used the stuff for what 2 decades or so, & ive heard this claim , & im aware of how conductivity can be dangerous to components.

its just that i get the 'wrist strap' feeling whenever i see someone post about/regarding the 'dangers' of AS5.
water is SUPER conductive, yet we cool electronics with it all the time, & there are lots of alternatives.

it just seems like this argument is a silly one to make (not that you personally were making an argument, im speaking more to the subject generally), nothing against you, its not like youre the first to post it, i just dont see the argument.

i dunno, maybe its just me

i didnt know as5 is conductive , anyway it never did anything wrong
 
Joined
Apr 9, 2018
Messages
771 (0.66/day)
I re-pasted my GTX 1060 with MX-5 and maximum temperature has gone from 78C down to 74C. From memory I'm pretty sure the old paste was GD900 that has been on there for a good 18 months.

Re-pasting the 5600X yielded less results, it was running GD900 for almost 6 months and it went from 78C down to 76C with MX-5. I'll take that, especially if it lasts the 8 years that they claim.

The MX-5 is definitely my new winner in terms of lower temperatures but damn the GD900 has held on well and removing the old paste was a cinch.
 
Joined
Mar 21, 2021
Messages
465 (5.54/day)
System Name HP Compaq 8000 Elite CMT
Processor Intel Core 2 Quad Q9550
Motherboard Hewlett-Packard 3647h
Memory 16GB DDR3
Video Card(s) NVIDIA GeForce GT 1030 GDDR5 (fan-less)
Storage 2TB Seagate Firecuda 3.5"
Display(s) Dell P2416D (2560 x 1440)
Power Supply 12V HP proprietary
Software Windows 10 Pro 64-bit
Joined
Jan 11, 2005
Messages
128 (0.02/day)
Kryonaut Extreme is a good paste. Probably 2C better than Thermalright TFX on my r9 290X. Although TFX was 2C better than regular Kryonaut on a 10900k after one week (started out like 1C worse).
The 290X reference heatsink has pretty decent and even pressure (enough to make a nice indentation in a graphite pad (IC / soft PGS+ (EYG-S), and EYG-R)), so it's no surprise Kryonaut Extreme would like those as Kryonaut always likes surfaces with even, firm mounting pressure. It's not as good with low mounting pressure though (as shown in the previous links). Also longevity is definitely an issue if the mounting pressure is not good enough, which I had with Kryonaut Extreme on my 3090 FE when I used Thermalright Odyssey 1.5mm pads on the core side. TFX lasted longer but even that degraded in a few weeks (core temp delta to core hotspot delta increasing!). Changing to Gelid Extreme 1.5mm pads, which are much softer, helped maintain much better control over the hotspot delta, but I haven't been able to test that with Kryonaut Extreme yet.

This whole thing really woke me up to how poorly designed the 3090 FE heatsink is. While it functions very well in heat movement, the design is actually atrocious. The only thing that screws the PCB to the heatsink when mounting the PCB on are the four leaf spring Torx screws. Since these are not spring loaded like the smaller X-bracket of the Radeon cards (the screws themselves are much better, unlike the "strip me" screws on the reference AMD heatsinks!!), this means that the the only thing pushing the GPU Core and PCB on is just that bracket! That means that there is absolutely no help with the memory pad compression whatsoever and that creates a major problem if you don't use the correct thermal pads, with a high softness. Yes, the backplate helps the edge of the PCB screw into the heatsink, but screws only on the very edge do absolutely nothing to the resistance caused by thermal pads surrounding the core!

Compare this to the AMD R9 290X reference. There you have like 12 screws all around the PCB, helping the PCB get very firmly secured all around the thermal pads of the VRAM and VRM's. Even though the Vega 64 has fewer screws there's still plenty to insure the VRAM pads get great contact. You get some really big indentions on Arctic thermal pad replacements (on Vega 64, a small section needs 1.5mm or 2mm on VRM's, the rest 1.0mm). If the FE actually had more screws, then there wouldn't be these problems at all. There really need to be 6 screws around the VRAM/VRM area of the 3090 FE / 3080 FE cards, just like AMD did it, then there would be no complaints at all.

Anyway,

MX-5 is a couple of C worse than Kryonaut / TFX on a desktop and a r9 290X, but it definitely should not be used with low mounting pressure (laptop video card may be fine but BGA CPU is probably asking for trouble, just like MX-4.. Kryonaut starts off well on any laptop and is great on the GPU, but pumps out fast on a CPU. Kryonaut Extreme is still really too new for these tests, but I already know from my 3090 FE that it doesn't do well with low pressure and a convex GPU core! (again I haven't tested it since switching to more compressible pads, which allowed better pressure on the core->heatsink).

 
Joined
Jul 5, 2013
Messages
13,461 (4.64/day)
Location
USA
System Name GPD-Q9
Processor Rockchip RK-3288 1.8ghz quad core
Motherboard GPD Q9_V6_150528
Cooling Passive
Memory 2GB DDR3
Video Card(s) Mali T764
Storage 16GB Samsung NAND
Display(s) IPS 1024x600
Joined
Apr 9, 2018
Messages
771 (0.66/day)
Don't forget to include diaper cream :) but only the good stuff (40% zinc oxide) and vaseline.

Price is a big factor
"An engineer can do for a dollar what any fool can do for two."
Arthur Mellen Wellington

OK, I couldn't take it no more... ordered me four 30g pots of GD900 ($3.51 a pot with shipping)

Now I find

GD32 1.0 W/K.m
GD33 1.0
GD66 1.05
GD100 1.094
GD220 1.114
GD280 1.159
GD380 2.2
GD460 2.0
GD450 2.0
GD900 4.8
GD900-1 6

I tried the newer GD900-1 version with the 6.0W/m2 thermal conductivity and it's pretty bad. It performs noticeably worse than the regular GD900.

GD900 Run 1 (Ambient 26.9C) = 77C, 74C, 80C, 78C
GD900 Run 2 (Ambient 26.8C) = 78C, 75C, 81C, 79C
GD900 Run 3 (Ambient 26.8C) = 79C, 76C, 82C, 80C

GD900-1 Run 1 (Ambient 27.0C) = 81C, 77C, 84C, 83C (NOTE: Core 2 thermal throttled at 84C)
GD900-1 Run 2 (Ambient 26.8C) = 80C, 77C, 84C, 82C (NOTE: Core 2 thermal throttled at 84C)
GD900-1 Run 3 (Ambient 26.6C) = 81C, 79C, 84C, 82C (NOTE: Core 2 thermal throttled at 84C)

Once again, it proves that the thermal conductivity value itself provided by thermal paste manufacturers means squat.

Of course I'm running the paste on both the die and IHS of my X5460 so any differences will be more pronounced, but it's still a huge difference. The GD900-1 paste is definitely more solid in consistency and harder to spread than the regular GD900. I have no doubt that the GD900-1 I received is the real product and not counterfeit. GD900 wins hands down and it's ridiculously cheaper, no competition.
 
Joined
Nov 11, 2016
Messages
1,519 (0.91/day)
System Name The de-ploughminator
Processor I7 9900K @ 5.1Ghz
Motherboard Gigabyte Z370 Gaming 5
Cooling Custom Watercooling
Memory 4x8GB G.Skill Trident Neo 3600mhz 15-15-15-30
Video Card(s) RTX 3090 + Bitspower WB
Storage Plextor 512GB nvme SSD
Display(s) LG OLED CX48"
Case Lian Li 011D Dynamic
Audio Device(s) Creative AE-5
Power Supply Corsair RM1000
Mouse Razor Viper Ultimate
Keyboard Corsair K75
Software Win10
I tried the newer GD900-1 version with the 6.0W/m2 thermal conductivity and it's pretty bad. It performs noticeably worse than the regular GD900.

GD900 Run 1 (Ambient 26.9C) = 77C, 74C, 80C, 78C
GD900 Run 2 (Ambient 26.8C) = 78C, 75C, 81C, 79C
GD900 Run 3 (Ambient 26.8C) = 79C, 76C, 82C, 80C

GD900-1 Run 1 (Ambient 27.0C) = 81C, 77C, 84C, 83C (NOTE: Core 2 thermal throttled at 84C)
GD900-1 Run 2 (Ambient 26.8C) = 80C, 77C, 84C, 82C (NOTE: Core 2 thermal throttled at 84C)
GD900-1 Run 3 (Ambient 26.6C) = 81C, 79C, 84C, 82C (NOTE: Core 2 thermal throttled at 84C)

Once again, it proves that the thermal conductivity value itself provided by thermal paste manufacturers means squat.

Of course I'm running the paste on both the die and IHS of my X5460 so any differences will be more pronounced, but it's still a huge difference. The GD900-1 paste is definitely more solid in consistency and harder to spread than the regular GD900. I have no doubt that the GD900-1 I received is the real product and not counterfeit. GD900 wins hands down and it's ridiculously cheaper, no competition.

Can you try again the GD900-1 and this time heat it up with a hairdryer before mounting the cooler? this would soften the paste and make it spread easier, create less resistance when tightening the screws --> less gap between CPU and cooler.
 
Joined
Jan 31, 2010
Messages
4,215 (1.02/day)
Location
Gougeland (NZ)
System Name Cumquat 2021
Processor AMD RyZen R7 3700X @4400MHz 1.325V
Motherboard Asus Strix x570-F Gaming
Cooling Cryorig R1 Universal
Memory 16GB GSkill TridentZ DR4-3200 16-18-18-38 1.35V
Video Card(s) Sapphire Nitro+ OC RX6800 16GB DDR6 1980Cclk / 2000Mclk
Storage 1x Adata SX8200PRO NVMe gen3 x4 1X Samsung 860 EVO 500GB, 12TB of HDD Storage
Display(s) ViewSonic VX2457 MHD 75Hz FreeSync 1920x1080p
Case SilverStone Raven RV02B-W + 3x SS AP181 + USB3.0 upgrade
Audio Device(s) RX6800 via HDMI + Pioneer VSX-531 amp Technics 100W 5.1 Speaker set
Power Supply Enermax RevolutionXT II 750W 80+ Gold
Mouse Logitech G502 Proteus Core
Keyboard Logitech G105
Software Windows 10 X64 PRO (build 20H2)
Benchmark Scores it sucks even more less now ;)
i didnt know as5 is conductive , anyway it never did anything wrong
It's not conductive at all although it is ever so slightly capacitive

im curious if anyone who actually should have their hands inside of a PC, has ever damaged something with Arctic Silver5. (rhetorical, im certain a million fools have damaged a million components all in equally ridiculous way)
ive used the stuff for what 2 decades or so, & ive heard this claim , & im aware of how conductivity can be dangerous to components.

its just that i get the 'wrist strap' feeling whenever i see someone post about/regarding the 'dangers' of AS5.
water is SUPER conductive, yet we cool electronics with it all the time, & there are lots of alternatives.

it just seems like this argument is a silly one to make (not that you personally were making an argument, im speaking more to the subject generally), nothing against you, its not like youre the first to post it, i just dont see the argument.

i dunno, maybe its just me

1: Artic Silver five is non conductive ( Although it is ever so slightly capacitive) so unless you use an entire tube of it you should never see a problem
2: Denatured water ie: water that has all of it's impurities removed so you're left with just pure H2O is also non conductive ( it's the impurities that make water conductive)
 
Joined
Jan 11, 2005
Messages
128 (0.02/day)
And why is this?

Because it's not thick and will end up pumping out due to natural heat related expansion / contraction cycles (you need a thick paste to fight against this).
The only paste that is good against imbalanced pressure like this is IC Diamond.

Naturally the best solution is to fix the mounting pressure or otherwise make the surfaces flat (sanding the heatsink, or if you have the courage, cpu/gpu die, if it's convex, etc)
 
Joined
Jul 5, 2013
Messages
13,461 (4.64/day)
Location
USA
System Name GPD-Q9
Processor Rockchip RK-3288 1.8ghz quad core
Motherboard GPD Q9_V6_150528
Cooling Passive
Memory 2GB DDR3
Video Card(s) Mali T764
Storage 16GB Samsung NAND
Display(s) IPS 1024x600
Because it's not thick and will end up pumping out due to natural heat related expansion / contraction cycles (you need a thick paste to fight against this).
The only paste that is good against imbalanced pressure like this is IC Diamond.

Naturally the best solution is to fix the mounting pressure or otherwise make the surfaces flat (sanding the heatsink, or if you have the courage, cpu/gpu die, if it's convex, etc)
I have never seen this with any paste, ever. I'm certain it's not going to be a problem, whether short or long term. Given how sticky MX-5 is, it seems very unlikely to flex and flow like you suggest.
 
Joined
Jan 11, 2005
Messages
128 (0.02/day)
I have never seen this with any paste, ever. I'm certain it's not going to be a problem, whether short or long term. Given how sticky MX-5 is, it seems very unlikely to flex and flow like you suggest.

Plenty of people have. Check notebookreview forums. This has been discussed to death already but countless people. If you bother to do a search, you'll see it all over. In the Alienware threads, MSI GT73VR threads, the thermal paste hardware discussion threads, it's all over.

And I've seen it on my 3090 FE when I used too hard (not compressible enough, even though they were correct thickness) thermal pads.
 
Joined
Jul 5, 2013
Messages
13,461 (4.64/day)
Location
USA
System Name GPD-Q9
Processor Rockchip RK-3288 1.8ghz quad core
Motherboard GPD Q9_V6_150528
Cooling Passive
Memory 2GB DDR3
Video Card(s) Mali T764
Storage 16GB Samsung NAND
Display(s) IPS 1024x600
Plenty of people have. Check notebookreview forums. This has been discussed to death already but countless people. If you bother to do a search, you'll see it all over. In the Alienware threads, MSI GT73VR threads, the thermal paste hardware discussion threads, it's all over.
I'm all over the place on the net and have never seen such a discussion. Got an example?
 
Joined
Dec 16, 2012
Messages
383 (0.12/day)
Processor AMD Ryzen R7 5800x
Motherboard B550i Aorus Pro AX
Cooling ID-Cooling SE234
Memory 32Gb Patriot Viper 3600 RGB
Video Card(s) MSI RTX 3080 Ventus Trio OC
Storage Samsung 960 EVO
Display(s) Specterpro 34uw100
Case Cooler Master NR200P
Power Supply Cooler Master V750 Gold SFX
VR HMD Rift S
Software Windows 10 64bit
The new Hardware Canucks video about Induim is interesting. The ihs imprinted on the indium pad and it performs like an mx4.
 
Joined
Oct 17, 2014
Messages
6,931 (2.85/day)
Location
USA
Processor i7-7820HK
Memory 2x8gb (16gb)
Video Card(s) gtx 1070
Storage 2TB SSD
Display(s) 17.3" 96hz 1080p IPS
The new Hardware Canucks video about Induim is interesting. The ihs imprinted on the indium pad and it performs like an mx4.

was this on a laptop or desktop?
 
Joined
Jan 11, 2005
Messages
128 (0.02/day)
I'm all over the place on the net and have never seen such a discussion. Got an example?
I'm sorry but I'm not going to search threads for you. You'll have to do that yourself.
Just put in the work and look on notebookreview. There's a big megathread on thermal paste in the general hardware section.
 
Joined
Jul 5, 2013
Messages
13,461 (4.64/day)
Location
USA
System Name GPD-Q9
Processor Rockchip RK-3288 1.8ghz quad core
Motherboard GPD Q9_V6_150528
Cooling Passive
Memory 2GB DDR3
Video Card(s) Mali T764
Storage 16GB Samsung NAND
Display(s) IPS 1024x600
I'm sorry but I'm not going to search threads for you. You'll have to do that yourself.
Ok, then. Not gonna bother. Been using and installing TIM products for decades and have never seen, read or heard about the behavior you described. I have some MX-5 and am not seeing it with that either. So you do you. The rest of us will carry on unabated..
was this on a laptop or desktop?
My guess is desktop as laptop & mobile CPU's don't have IHS plates.
 
Joined
Jun 3, 2010
Messages
1,914 (0.48/day)
Because it's not thick and will end up pumping out due to natural heat related expansion / contraction cycles (you need a thick paste to fight against this).
The only paste that is good against imbalanced pressure like this is IC Diamond.

Naturally the best solution is to fix the mounting pressure or otherwise make the surfaces flat (sanding the heatsink, or if you have the courage, cpu/gpu die, if it's convex, etc)
Isn't it the other way around, IC Diamond drying out over time?
OP: your thread is a huge let down. Do review in the future when you say so.
 
Top