[Solved] MSI R9 390x Overheating (possible related to VRM)

[Slodin]

Hi,

 

I was experiencing overheating problems for a while now from my MSI R9 390x, it started strong being able to handle 80-92 degrees when it was purchased in the late 2015. However, about last year it started to overheat around 85ish and now it overheats at mere 67. I monitored the temperature for a bit and realized it's probably not the core that is overheating, but possibly the VRM.

 

The card has 2 VRM sensors, VRM1 reads normally and maxing around 50 degrees but VRM2 is stuck at a number when the machine is turned on. As soon as the pc is turned on, the reading of VRM2 is between the early 40s and early 70s and never changes until I shut the pc off (restarts don't work either). VRM1 on the other hand starts at 19 and slowly work it's way up to around 27 at idle, it will climb to around late 40s when on full load. 

 

Somehow I can hardly find much information regarding gpu VRM issues, specially the sensor being stuck. I can only suspect the VRM2 is being the problem here causing all the freeze ups and game crashes.

I ordered some thermal pads and new paste hoping for a fix.

I will include a hwinfo log as attachment, I would really appreciate if someone can help me out.

Log.CSV

 

UPDATE 9/27/2018:

• Was having blue lines randomly appearing on desktop so I thought it's dead as pointed out by a mate on this forum.
• Contacted MSI for a RMA and got refused because of serial number saying the date is July 2015 and now it has been over 3 years thus warranty is expired. Yeah so much for warranty when it fails 2 months after.
• Already bought new thermal pads and thermal paste and since warranty expired it doesn't hurt to tear the whole thing up.
• Found the problem, all the pads are dry and crumbling due to old age. A few memory chips are literally half covered by MSI's craftsmanship, and the GPU core itself is half covered in thermal paste too. Nicely done MSI!
• Replaced all pads and paste, now running with no problem at the moment. (Few hours of gaming at 80 degrees, it's actually hard to reach 80 at this point with everything replaced)

I don't expect it to live long tho...because of all the failures from before might have damaged it and shorten the lifespan. Since it's a SOL card, might as well learn something.

 

 

[Jurrunio]

Yeah but what model is this?

[Slodin]

1 minute ago, Jurrunio said:

Yeah but what model is this?

oh shit, sorry. I changed the title.

It's MSI

[LamoidZombieDog]

1 minute ago, Slodin said:

oh shit, sorry. I changed the title.

It's MSI

MSI R9 390x You mean

[Slodin]

2 minutes ago, LamoidZombieDog said:

MSI R9 390x You mean

yes, that's what I changed the title to

[Jurrunio]

16 minutes ago, Slodin said:

oh shit, sorry. I changed the title.

It's MSI

That's just a brand, not model. Those VRM temps are wayyyyy off for a reference PCB design or air cooled cards.

[AlexWarFire]

Maybe clean it up and replace thermal paste, but if you're not comfortable, just get a new graphics card.

[Slodin]

3 minutes ago, Jurrunio said:

That's just a brand, not model. Those VRM temps are wayyyyy off for a reference PCB design or air cooled cards.

Honestly, I thought the model is R9 390x, so I'm just going to leave this copied from my AMD panel here with device IDs and stuff...if this is useful.

 

Graphics Card Manufacturer - Powered by AMD
Graphics Chipset - AMD Radeon (TM) R9 390 Series
Device ID - 67B0
Vendor ID - 1002
SubSystem ID - 2015
SubSystem Vendor ID - 1462
Revision ID - 80
Bus Type - PCI Express 3.0
Current Bus Settings - PCI Express 3.0 x16
BIOS Version - 015.048.000.062
BIOS Part Number - MS-V30823-F2
BIOS Date - 2015/05/21 03:39
Memory Size - 8192 MB
Memory Type - GDDR5
Memory Clock - 1500 MHz
Core Clock - 1080 MHz
Total Memory Bandwidth - 384 GByte/s
Memory Bit Rate - 6.00 Gbps
 

[Tristerin]

1 minute ago, AlexWarFire said:

Maybe clean it up and replace thermal paste, but if you're not comfortable, just get a new graphics card.

This would be step 1, and likely to cure the issue.  Or, you should place fans directly on it (open your box) and see if that helps.  (Zip tie some smaller case fans around it if you are to uncomfortable as said above)

[aezakmi]

Quote

it started to overheat around 85ish and now it overheats at mere 67

that doesn't makes any sense, AFAIK 67 is lower than 85 so there's something wrong here

 

either you call overheat to something else or the temps are inverted being 67 the temp when you bought the card and 85 now

 

 

[Slodin]

4 minutes ago, AlexWarFire said:

Maybe clean it up and replace thermal paste, but if you're not comfortable, just get a new graphics card.

 

1 minute ago, Tristerin said:

This would be step 1, and likely to cure the issue.  Or, you should place fans directly on it (open your box) and see if that helps.  (Zip tie some smaller case fans around it if you are to uncomfortable as said above)

 

yeah the replace paste and thermal pad is what I'm going to do...Just didn't know if this would be the fix or something else is wrong with the card. I'm pretty comfortable in changing those.

[AlexWarFire]

It's your choice, but it will definitely void your warranty, and try to not break it

[Jurrunio]

So the model name is MsI R9 390X Gaming. The custom air cooled card with two fans.

 

That VRM sensor of some sort's number isnt realistic though, VRMs powering an R9 390X at stock should easily hit 80C, topping at about 90C. Maybe you're rading the wrong number?

[Slodin]

1 minute ago, aezakmi said:

that doesn't makes any sense, AFAIK 67 is lower than 85 so there's something wrong here

 

either you call overheat to something else or the temps are inverted being 67 the temp when you bought the card and 85 now

 

 

That's exactly why it's so weird, and took me awhile to grasp. I have been monitoring the temps for a while now and that's what I noticed. Thus I moved my monitoring to the VRM rather than the core temp now because I feel like it's not the overall GPU overheating, but VRM2.

[Slodin]

3 minutes ago, Jurrunio said:

So the model name is MsI R9 390X Gaming. The custom air cooled card with two fans.

 

That VRM sensor of some sort's number isnt realistic though, VRMs powering an R9 390X at stock should easily hit 80C, topping at about 90C. Maybe you're rading the wrong number?

I thought there is only one model of the 390x MSI produces...maybe i'm wrong. But yeah, air cooled with two fans and a back plate.

I don't think I'm reading it wrong as it says from HWINFO64: GPU VRM Temperature 1, GPU VRM Temperature 2

 

yeah, it isn't very realistic because I cannot get it to run to a stable temperature since it just crashes. The Log CSV i provided in the main post have both VRM temps from start of machine to crash. Basically, start VRM1 = 33, VRM2 = 69. End VRM1 = 44, VRM = 69. The second VRM never changed, which I can only assume VRM2 is overheating since it starts off really high.

[aezakmi]

1 minute ago, Jurrunio said:

So the model name is MsI R9 390X Gaming. The custom air cooled card with two fans.

 

That VRM sensor of some sort's number isnt realistic though, VRMs powering an R9 390X at stock should easily hit 80C, topping at about 90C. Maybe you're rading the wrong number?

this

 

1 minute ago, Slodin said:

t's not the overall GPU overheating, but VRM2.

IF 67C is the correct temp here then that's not hot AT ALL for a 390 or 390X

my old 290X could easily hit 90C while gaming probably because the MSI heatsink is kinda crappy being too small for such card, I did replaced a few things though, used regular MX4 for the chip and AS Matrix for VRMs, lowered temps to about 75C under the same loads without oc, added a fan too

[Jurrunio]

5 minutes ago, Slodin said:

I thought there is only one model of the 390x MSI produces...maybe i'm wrong. But yeah, air cooled with two fans and a back plate.

I don't think I'm reading it wrong as it says from HWINFO64: GPU VRM Temperature 1, GPU VRM Temperature 2

 

yeah, it isn't very realistic because I cannot get it to run to a stable temperature since it just crashes. The Log CSV i provided in the main post have both VRM temps from start of machine to crash. Basically, start VRM1 = 33, VRM2 = 69. End VRM1 = 44, VRM = 69. The second VRM never changed, which I can only assume VRM2 is overheating since it starts off really high.

1st VRM tenperature could be memory power, considering how little GDDR5 power draw changes under different situations.

 

2nd VRM temps then will definitely be GPU power, of ehich its readings still doesnt corresspond to how a 390X should perform.

 

My guess is that the crashes originate from GPU instability at this voltage. Try push the clock speed down or the voltage up, see what happens. Also, if the power limit is being hit, raise that as well.

[Slodin]

Just now, aezakmi said:

this

 

IF 67C is the correct temp here then that's not hot AT ALL for a 390 or 390X

my old 290X could easily hit 90C while gaming probably because the MSI heatsink is kinda crappy being too small for such card, I did replaced a few things though, used regular MX4 for the chip and AS Matrix for VRMs, lowered temps to about 75C under the same loads without oc, added a fan too

I know the 390 is suppose to take a lot of heat because I had it for a few years and knows that...

 

The VRM2 temperature never changes as in the sensor is stuck at that reading..forever (until I power the pc off). Even if the actual temperature changes the reading doesn't, since it doesn't make any sense for it to begin(start of pc) at 69, idle for 10 mins still at 69 then load a game for 5 mins and still be at 69 with no changes whatsoever. Of course, the game crashes in a few minutes too...

[Slodin]

57 minutes ago, Jurrunio said:

1st VRM tenperature could be memory power, considering how little GDDR5 power draw changes under different situations.

 

2nd VRM temps then will definitely be GPU power, of ehich its readings still doesnt corresspond to how a 390X should perform.

 

My guess is that the crashes originate from GPU instability at this voltage. Try push the clock speed down or the voltage up, see what happens. Also, if the power limit is being hit, raise that as well.

I was going to try your suggestion and before I even got a chance, it started acting up even on windows desktop. Artifacts and blue lines started to appear randomly with system crashes...Checked warranty and noticed the card is 2 month out of the warranty period.... I just shut the system off and on my laptop now...at this point I'm not sure if I can fix it.

[Jurrunio]

2 minutes ago, Slodin said:

I was going to try your suggestion and before I even got a chance, it started acting up even on windows desktop. Artifacts and blue lines started to appear randomly with system crashes...Checked warranty and noticed the card is 2 month out of the warranty period.... I just shut the system off and on my laptop now...at this point I'm not sure if I can fix it.

Artifacts is usually caused by failing memory modules or loose memory modules (solder coming off). Dragging memory clock all the way down could help stability after boot. During boot though it will still hit high clocks, potentially crashing things

 

Then it's apparently on its last legs. You can ask MSI if they can fix it, but I dont think its worth fixing.

[Slodin]

2 minutes ago, Jurrunio said:

Artifacts is usually caused by failing memory modules or loose memory modules (solder coming off). Dragging memory clock all the way down could help stability after boot. During boot though it will still hit high clocks, potentially crashing things

 

Then it's apparently on its last legs. You can ask MSI if they can fix it, but I dont think its worth fixing.

Yeah I contacted MSI just now. Worse I get is a no for RMA...

This problem is solved so fast...cause most likely...new card...

[Slodin]

Updated post with my solution...May it help someone that is running into the same problem as I did.