Bitcoin Forum

I have two MSI Armor cards, one runs at 75C/75% fan or below dual mining Eth and Dcr, one thermal throttles at 85-88C (reported by hwinfo/trixx) mining just Eth, since I manually disabled Dcr mining as the heat is a problem. Interestingly, the hot problematic one is like 88% asic, and undervolted to 850mV core/940mV mem controller (confirmed w/hwinfo) running at 1150 max boost (which it never reaches). Power draw is also low, around 60-70w VRM in. The fan is set to 100%, and it is spinning at 3200+RPM. Thinking it was a shitty thermal paste job, I repasted it but the temps are still shit. Card is less than 2 weeks old and has been running hot from the start. Any ideas why the temp is apparently so high? Thanks in advance.

Edit: The cards are at least 2cm apart, the same as my 9 other 470s that are all running well below 80C (there is a big box fan sucking in sub freezing air from outside behind my rig as well)

try to swap the entire cooler  with the good msi armor  see if the problem follows the cooler.


Also run only the 2 msi cards on a rig  with a lot of room to breathe.


 are they sealed new from a store or used from some guy

maybe they had a rom flash


Lastly check what ram they have   gpu-z  will show you.

Just an opinion but MSI's quality control isn't the best. You may just have a poorly made card.

If you have the ability to do so, I'd just exchange it with your place of purchase and see if you get something that performs better. I have an XFX card in 1 rig that acts similar to yours and I wish I would have just exchanged it when I had the chance.

Now many months later, I'm stuck with its hot tempered nature.

It is true  that bad cards do exist   so you could be right.
I have 29 of these  they are pretty good.  But one drops  msi settings on afterburner every once in a while

http://wimages.vr-zone.net/2016/08/MSI-RX-470-Gaming-X.jpg




Op has these I do not know if cooler is  the same as mine.


https://cdn.videocardz.com/1/2016/08/MSI-RX-470-1.jpg

For reference: the RX470 Armor is a different card entirely than the RX470 Gaming

I currently have 7 of the RX470 Gaming GPU's and was going to buy some of the Armor before I saw the difference. I don't care for the Armors design.

470 Armor 4gb
https://s5.postimg.org/3krbtyujn/470_Armor.jpg (https://postimg.org/image/3krbtyujn/)

470 Gaming 4gb
https://s5.postimg.org/rd0n5hwkj/470_Gaming.jpg (https://postimg.org/image/rd0n5hwkj/)

Yeah, gaming x is much better. Have a couple myself. The hot armor still isn't fixed yet, but I ordered a dead non reference 7970 for cheap so I will swap that cooler onto the rx470 if it fits (should be fine).

Thanks for this!

I know it's an old thread (found searching on google) but I had a couple older cards laying around that for some reason ran extremely hot no matter what I did. I tried the usual things like reapplying thermal compound, cleaning dust out, running super low voltages, etc.

After finding this I swapped the entire heat-sink cooler with another that ran fine and it worked. Until coming across this post I would have never thought the cooler would be the problem, as its mainly a hunk of metal with two fans attached, which in my case the fans were good.

So is it the liquid inside the little tubes that leaks out or what? I might try my luck on eBay and buy some parts-only GPUs for the cooler portion, but how to know if they are good?

yes might be liquid inside copper pipe leaked, but i dont think cause to much for rx 470 card which is low power gpu, might be some bent in mounting and die not connecting evenly in heatsink
this happen on many vegas gpu. trying to replace with new heatsink to costly for gpu already fall in price

In 2018 i had rx470 armor like authors of this post. You are right heat pipe was not working. You can easily diagnouse it by touching the the tubes or if you have couple of multimeters it has temperature sensor. Attach the sensor on tube and look.in my case heat pipe was not reacting at all. I replaced radiator from dead xfx 480. I need to cut some parts but it is still working this days.

I have rescued so many gpus by heatsink switching sometimes adding pads sometimes adding huge fans.

lots of rescue methods.

I am older and my eyes are not as good as they were back around 2005 or so. so I do not try solder fixing.

but playing around with heatsinks I can still do.

I am putting a ton of gpus back online I am back to 100 plus cards.

power is cheap so I am making decent money again.

In my case it was a RX580, but same series basically.

I did notice on this particular model (MSI Armor) that when you take the heat sink off the GPU, on the backside there is a small copper tube inset in the block with the two aluminum tubes connected one each end. I wonder if this is why this particular brand/model leaks more. I have other similar GPUs from that era that have a similar design that use solid aluminum throughout with the piping (tubes) all being one piece instead of two.

Anyway, I assume that what ever fluid they use inside these tubes, or heat pipes, boiled away over time and now they are just empty tubes of air acting more like an insulator than a conductor of heat. All that is left is the direct metal contact of the remaining portion of the heat sink which will transfer some but not all of the heat away.

Well thanks again for sharing your knowledge.

I had a bin of old GPUs that no longer worked for whatever reason, some didn't run at all, others had high heat issues. I have swapped fans and replaced thermal paste and pads before, but the idea of the metal heat-block being bad was one that never occurred to me. Well now I managed to save a couple more from the landfill.

MSI Armor is a reliable video card in terms of elements, but it has very poor fans and cooling radiator. If there is overheating, it is usually due to gas escaping from the radiator tubes.
On the secondary market, these radiators are bad and they cost 80% or more of the cost of the used video card.  My video cards mine the Kawpow algorithm with a temperature of 90 degrees with a bad heatsink and they are still alive.

Hi FP91
Nice to see people still here and talking.
Hows your rigs running?
I have armor rx 470 running since 2018 with original fans  8) fans quality amazing they just dont die.i have 3 armor rx 470 and one armor 480 8gb. I think they had problem in factory with heat pipes for this model thats why i saw same problems with armor rx470. Worst fans is xfx rx570.red fans

yeah xfx were always troublesome.

Question on the rx 470 cards I have six maybe seven blower cards.

what do they give if mining Kawpow.. I have a six card board I can set up do they do 16 or 18 or better for kawpow?  i think watts are around 120.

what should core be?

i use simple mining.

core 1150?
volts.  .700

mem.    900

volts 1.300
vots.   .750

have not run them in a while.

Core   1150 MHz
Memory   2000 MHz I had problems when increasing the frequency, so 2000 is optimal. But you can overclock to 2100, although on Kawpow there will not be a big increase in hashrate, and the memory will overheat.
volt   875-900 mv     This is not Ethereum, Kawpow loves to consume electricity


Over the past 5 years, all the fans on my MSI Armor have died. I bought Chinese analogues and sold video cards. I left video cards with bad heatsinks for my own needs. Chinese fans die within 2-3 months of mining, so I use 80x80.

Your numbers way good than mine 16-18. I get 13-14 using teamred but i lower power. 1125 core 2000 memory and voltage same for both 890mv. And there is trick i still overclock them using phoenix miner. I give all oc settings in bat file and write -leaveoc.phoenix miner has -vmr parametr which gives extra hashes. Claymore gave name for this rxboost. It memory refresh rates.i use -vmr 30
There is bench mode on phoenix miner.when you write -bench 50 it bench your rig for epoch 50 and applies all oc setting to cards  8)

The profit on the Kawpow algorithm has already fallen because miners started using old video cards. For the RX470, I would choose a mode with more stable operation, because any memory overclocking on my video cards forced me to periodically reboot the mining rig remotely. That's why I only configured the core/