[SOLVED] I bought MSI X570-A PRO for Ryzen 9 3900X. It has bad VRM. Should I be worried?

Nov 20, 2019
3
0
10
0
I bought this motherboard not because I could not afford anything more expensive, but because according to the reviews I was able to find at the time of purchase it should be good enough for 12 core CPU from AMD. I don't care about any fancy features, I just care about overall system stability. And I am certainly not going to overclock anything.
But later I found this review from Hardware Unboxed
View: https://www.youtube.com/watch?v=xbyWKufthS4
which says that this motherboard has really bad VRM which easily overheats under load in certain conditions, and this forum seems to confirm it.
So far I did not have any problems, my PC seems to be working just well. I think it helps that I don't stress my system too much, and that the case fans are blowing directly at the VRM heat sinks.
However, I can afford to replace this motherboard with something better, and the prime candidate is ASUS Prime X570-P.
Should I waste few more hours rebuilding my PC on a new motherboard?
 
....
Should I waste few more hours rebuilding my PC on a new motherboard?
Sure, it would have been good to avoid the board in the first place. But, extreme tests to demonstrate a point aside, what's important is how it is performing in your useage. Since you got it, that is.

It's important to know what the VRM temperatures are in your normal useage scenarios. You can check that with a utility called HWInfo64, it should have a readout for VRM temp. It may be called something like 'System' or the like. To find it, put the system under load and watch which temperatures are rising up.

Once you've identified it go do what you normally do, monitoring that temperature occasionally. Please keep in mind that running stress tests and CPU benchmarks, while useful for the purpose intended, isn't really 'normal useage'. If it's not getting really hot in true normal useage, you'll be just fine.

Most people really do not hit the CPU with hard enough processing loads to raise the temperature enough to matter. Gaming, for instance, really doesn't do it. You may very well find that's your case.

Otherwise, simply on principle you could take it apart and replace it. That's your choice.
 
Last edited:
Reactions: higgs_boson
....
Should I waste few more hours rebuilding my PC on a new motherboard?
Sure, it would have been good to avoid the board in the first place. But, extreme tests to demonstrate a point aside, what's important is how it is performing in your useage. Since you got it, that is.

It's important to know what the VRM temperatures are in your normal useage scenarios. You can check that with a utility called HWInfo64, it should have a readout for VRM temp. It may be called something like 'System' or the like. To find it, put the system under load and watch which temperatures are rising up.

Once you've identified it go do what you normally do, monitoring that temperature occasionally. Please keep in mind that running stress tests and CPU benchmarks, while useful for the purpose intended, isn't really 'normal useage'. If it's not getting really hot in true normal useage, you'll be just fine.

Most people really do not hit the CPU with hard enough processing loads to raise the temperature enough to matter. Gaming, for instance, really doesn't do it. You may very well find that's your case.

Otherwise, simply on principle you could take it apart and replace it. That's your choice.
 
Last edited:
Reactions: higgs_boson
Nov 20, 2019
3
0
10
0
You can check that with a utility called HWInfo64, it should have a readout for VRM temp
That's a really good tip, thanks.

My primary use case is compiling on Linux which can fully load all cores for many minutes. But since HWInfo64 is only available on Windows, I ran Cinebench for 10+ minutes, it should simulate my workload pretty well. HWInfo64 has lots of sensors, I think the one that I am looking for is VR MOS. I could get it as hot as 82 degrees Celsius. Is this a normal temp?

Otherwise, simply on principle you could take it apart and replace it. That's your choice.
I think it makes sense to replace both MB and stock cooler. The stock cooler is quite audible at times.
 
That's a really good tip, thanks.

My primary use case is compiling on Linux which can fully load all cores for many minutes. But since HWInfo64 is only available on Windows, I ran Cinebench for 10+ minutes, it should simulate my workload pretty well. HWInfo64 has lots of sensors, I think the one that I am looking for is VR MOS. I could get it as hot as 82 degrees Celsius. Is this a normal temp?
...
82 C, while hot, isn't excessively so for the FET's used in VRM's. They are rated with operating temperatures north of 125C. But they will usually impose throttling on the CPU at 105-110C.

I've not worked with Linux compiling, not really sure what it entails instructions-wise. But don't confuse 'loading up the cores' with being a heavy workload. Most workloads aren't nearly as heavy as Cinebench with it's use of AVX instruction set for image rendering. A MalwareBytes virus scan, by comparison, also loads up the cores but the CPU hardly breaks a (thermal) sweat on my system while CB gets the fans going as temps rise.
 
Reactions: higgs_boson

jon96789

Upstanding
Aug 17, 2019
334
38
240
8
VRM temps in the 80 degree range is acceptable. It's when they exceed 90 degrees you may run into issues. My MSI MPG X570 Gaming Pro Carbon with the AMD 3900X would hit 95+ degrees when encoding videos. That's when it throttled the CPU speed down until the VRMs cooled down.

Apparently your compiling is not a high enough load to increase the temps into throttle limits. Looking at HWiNFO64, what percentage is the cores running? Is it under 100%? If so, the CPU is not under full load. Also, you did not say which CPU you are using. Is it a 65-watt or a 105-watt? Also, replacing the stock cooler will not make that much of a difference with the VRM temps.

When I am encoding videos, the CPU load is usually 95-100%. Althogh my AMD 3900X is supposed to be able to hit 4.6GHz, I find that the fastest the CPU will get is about an average of 4.0 GHz on all cores with couple of cores peaking at 4.5 GHz at instantaneous intervals.

I ended up buying the ASUS X570 ROG Crosshair VIII Hero, a board tested with an excellent VRM design. My thermals dropped by 50-60 degrees C under load. This board sells for $360, but the ASUS Prime X570-P board ($200) has excellent VRM temps for the price, testing at about 70 degrees with a 3900X under load. Just make sure that the board meets all of your requirements before buying.
 
Reactions: higgs_boson
Nov 20, 2019
3
0
10
0
VRM temps in the 80 degree range is acceptable. It's when they exceed 90 degrees you may run into issues.
I just want to be sure that I have a robust system. This is the first time I decided to build a PC by myself. I very much liked it, although it seems like I will have to pay for the lack of experience and a wrong choice.

Also, you did not say which CPU you are using. Is it a 65-watt or a 105-watt?
It's in the title of the topic. I have the same CPU as yours, Ryzen 9 3900X. It compiles like a monster ;)

Also, replacing the stock cooler will not make that much of a difference with the VRM temps.
Yes, I am aware of this. I just want the whole system to be more silent, that's all.

I ended up buying the ASUS X570 ROG Crosshair VIII Hero, a board tested with an excellent VRM design. My thermals dropped by 50-60 degrees C under load. This board sells for $360, but the ASUS Prime X570-P board ($200) has excellent VRM temps for the price, testing at about 70 degrees with a 3900X under load. Just make sure that the board meets all of your requirements before buying.
That's good to know, thanks. As I said, I don't want any "gaming" features, like RGB, overclocking or whatever. All I want is a basic but robust motherboard.
 

ASK THE COMMUNITY

TRENDING THREADS