Question Are MSI X.570 motherboard's VRMs really that bad?

jon96789

Reputable
Aug 17, 2019
414
49
4,740
I have been seeing videos on YouTube about MSI's X.570 motherboards having really crappy VRMs... Other than the TOTL Godlike, Hardware Unboxed reviewed a couple of MSI boards and (MSI X.570A Pro and the MSI X.570 Gaming Edge) and their tests show that the VRM design is really bad. They show the VRM temps for both boards as being over 100-125 degree Celsius (which is even higher than the 3900X CPU), where other boards run in the 60-80 degree range. They noted that these two boards have 50% less VRM components than the Godlike which tested in the 50-60 degree range.

Because of the high temps, the 3900X had throttled back on their speed until the VRMs recovered. The VRMs reached high temps and again the CPU throttled back and this cycle was repeated.

How does this relate in real world situations (I believe they removed the heat sinks for testing on the VRMs but not sure)? I have a MSI MPG Gaming Pro Carbon and am wondering if this board suffers the same problem...
 
  • Like
Reactions: brokeBuilder2019

Phaaze88

Titan
Ambassador
I know - and saw - exactly which videos you're talking about. It's just X570(currently) that they dropped the ball on.
Ironically, their B450 boards were well received.

It took 1 minute for it to throttle, and another minute for the temps to cool down and start rising back up again.
Steve did not remove the heatsinks. The tests were run on the open test bench, and depending on airflow through the individual case, expect temps to be slightly better or worse than his.

Your motherboard: https://www.tomshardware.com/reviews/msi-mpg-z390-gaming-pro-carbon-9900k-atx-motherboard,5856.html
Nope, it doesn't.
 
  • Like
Reactions: brokeBuilder2019

jon96789

Reputable
Aug 17, 2019
414
49
4,740
The layout of the X570 VRMs are different from the Z390 VRMs... The Z390 has 11 VRMs and the X570 only has 10...

OC3D.NET showed that the Pro Gaming Carbon X399 was about 20 degrees hotter than the ASUS ROG...

08085940476l.jpg
 
Last edited:

jon96789

Reputable
Aug 17, 2019
414
49
4,740
Found the specs on the net...

MEG X570 GODLIKE 18-phase[14+4]*| TDA21472 70A, PWM: IR35201(8-phase), PCB: ?-Layer
PRESTIGE X570 CREATION 16-phase[14+2]*| IR3555 60A, PWM: IR35201(8-phase), PCB: ?-Layer
MEG X570 ACE 14-phase[12+2]*| IR3555 60A, PWM: IR35201(8-phase), PCB: ?-Layer
MPG X570 GAMING PRO CARBON WIFI 12-phase[10+2]*| UBIQ QA3111N6N 56A, PWM: IR35201(8-phase), PCB: ?-Layer
MPG X570 GAMING EDGE WIFI 10-phase[8+2]*| OnSemi NTMFS4C029NT1G/NMMFS4C024NT1G 46A/78A, PWM: IR35201(8-phase), PCB: ?-Layer
MPG X570 GAMING PLUS 10-phase[8+2]*| OnSemi NTMFS4C029NT1G/NMMFS4C024NT1G 46A/78A, PWM: IR35201(8-phase), PCB: ?-Layer
MPG X570-A PRO 10-phase[8+2]*| OnSemi NTMFS4C029NT1G/NMMFS4C024NT1G 46A/78A, PWM: IR35201(8-phase), PCB: ?-Layer
 

Phaaze88

Titan
Ambassador
The layout of the X570 VRMs are different from the Z390 VRMs... The Z390 has 11 VRMs and the X570 only has 10...

OC3D.NET showed that the Pro Gaming Carbon X399 was about 20 degrees hotter than the ASUS ROG...

large chart
I'm aware the layout is different.
The Z390 having 1 more VRM over the X570 means what, exactly? Numbers aren't everything.
That result isn't terrible at all compared to what's going on over at X570. They made 2 other X399 boards so how did they fare?

Found the specs on the net...

MEG X570 GODLIKE 18-phase[14+4]*| TDA21472 70A, PWM: IR35201(8-phase), PCB: ?-Layer
PRESTIGE X570 CREATION 16-phase[14+2]*| IR3555 60A, PWM: IR35201(8-phase), PCB: ?-Layer
MEG X570 ACE 14-phase[12+2]*| IR3555 60A, PWM: IR35201(8-phase), PCB: ?-Layer
MPG X570 GAMING PRO CARBON WIFI 12-phase[10+2]*| UBIQ QA3111N6N 56A, PWM: IR35201(8-phase), PCB: ?-Layer
MPG X570 GAMING EDGE WIFI 10-phase[8+2]*| OnSemi NTMFS4C029NT1G/NMMFS4C024NT1G 46A/78A, PWM: IR35201(8-phase), PCB: ?-Layer
MPG X570 GAMING PLUS 10-phase[8+2]*| OnSemi NTMFS4C029NT1G/NMMFS4C024NT1G 46A/78A, PWM: IR35201(8-phase), PCB: ?-Layer
MPG X570-A PRO 10-phase[8+2]*| OnSemi NTMFS4C029NT1G/NMMFS4C024NT1G 46A/78A, PWM: IR35201(8-phase), PCB: ?-Layer
All I see are power phases that could even handle Threadrippers - and possibly the new ones - if AMD actually made them compatible with these boards.
The power delivery is so overkill for normal Ryzens...
Of course, none of this takes into account the VRM cooling solution. Make a cheap heatsink, expect to pay the price in... temps?

Look at the heatsinks of the boards(high-end price range) in this video and guess which ones do the worst of the 5 featured:
View: https://www.youtube.com/watch?v=gOSFX2tHDcE
 
Last edited:
  • Like
Reactions: WildCard999
I know - and saw - exactly which videos you're talking about. It's just X570(currently) that they dropped the ball on.
Ironically, their B450 boards were well received.
...

It's really sad... MSI put really good VRM's on B450 and X470 when both Asus and Gigabyte came out with some genuine crap for B450 and X470 (below their super-premium line). But for X570 they all three did a complete180: Asus and Gigabyte's VRM's are generally overkill even for a 3900x while MSI...well...watch that review.

Really sad...and makes my kinda wonder what MSi were thinking. I remember an interview Steve (GamersNexus) with MSI's CEO where he kept emphasizing that AMD wanted to go 'super premium' with the X570 line and it was going to be expensive. Then they deliver this?
 

Phaaze88

Titan
Ambassador
It's really sad... MSI put really good VRM's on B450 and X470 when both Asus and Gigabyte came out with some genuine crap for B450 and X470 (below their super-premium line). But for X570 they all three did a complete180: Asus and Gigabyte's VRM's are generally overkill even for a 3900x while MSI...well...watch that review.

Really sad...and makes my kinda wonder what MSi were thinking. I remember an interview Steve (GamersNexus) with MSI's CEO where he kept emphasizing that AMD wanted to go 'super premium' with the X570 line and it was going to be expensive. Then they deliver this?
It's greed. They cut corners at one of the most important parts of a motherboard so they could pocket more money for themselves.
Without question, as a company, you're going to sell more cheaper/budget mobos than high-end ones. They tried to play dirty, and got caught.

While not directly related, also consider the praise Msi usually gets for the crazy high memory overclocks on their boards.
I can't remember if it was X470, Z390, or whatever, they boasted about one of their mobos being able to hit 5500mhz. It's not practical, but it's there.
 

jon96789

Reputable
Aug 17, 2019
414
49
4,740
I decided to run Prime95 on my PC today to get an idea just how bad the MSi boards really are... The power setting in Control Panel was set to AMD Ryzen Performance... My MSi MPG X570 Gaming Pro Carbon has a simple 5+2 VRM design which slightly better than the bottom three MSi boards which only have 4+2 VRM design. According to Hardware Unboxed, the 4+2 VRM boards hit over 110 degrees C (up to 50 degrees hotter than some other boards), so I expected that the 5+2 design would be slightly cooler. They said that the board would be fine with AMD's 65 watt CPUs because of their lower power draw, but recommended users to avoid the boards for the 95 watt AMD CPUs or users that oc'ed their CPUs.

After 20 minutes in an air-conditioned environment, the stats are as follows:

CPU Temp peaked at 97 degrees C, averaged about 90 degrees...
CPU voltage was a constant 1.487 volts
CPU speed peaked at 4.45 GHz (far short of the rated 4.6 GHz) and averaged 4.2 GHz

The VRM ran at a whopping constant 95 degrees C according to HWInfo64 so it was slightly cooler as expected but still an excessively higher temp than recommended... Oddly, using an infrared laser temp gun at the VRMs (well, on the cubed components on the board by the heat sinks) registered 80 degrees. I don't know how accurate these IR guns are. After terminating Prime95, it took about 10 minutes before the VRM temps started dropping. I think I can safely assume that the board should not thermal throttle the 3900X like the 4+2 boards do.

I noticed at that time, my Corsair H115i RGB Platinum water block was acting up, the LEDs were flashing off and on iso running a rainbow scrolling pattern. This issue remained until the CPU temps went down (the AIO was running in balanced mode).

This shows how MSi really cut corners on the X570 VRM design. It is no wonder that MSi does not give the specs for the VRMs on their low-end boards but prominently displays them on the higher priced boards.
 
  • Like
Reactions: brokeBuilder2019

Phaaze88

Titan
Ambassador
1)Those temps are with the latest ABBA bios update? That's nuts!

2)VRM cooling is a potential weakness of liquid cooling cpus. Having a motherboard with an already weak VRM heatsink doesn't help things.
Air cooled setups sufficiently cool the VRMs via the combination of top-down, or tower cpu cooler and top case fan exhaust.
 
I decided to run Prime95 ...

Prime 95 is a power virus that in no way accurately reflects any kind of real-world work load. I'd draw no conclusions running that alone. In fact, 95C on VRM running P95 could actually be indicator of pretty good rubustness when Vmax on the FET's is often up around 125C.

How does it perform running RealBench? What temp reading are you getting then? That's real-world, heavy work load that, while may be rare in practice reflects what's actually possible.

Even so, I'm not trying to say MSI's X570 boards have great VRM's even if they are good enough to do the job. The issue with them is that X570 boards are too expensive and for that I think it's fair to look for top-line value everywhere in the board, not just a PCIE gen 4 chipset. When you can get that in even lower-end models from Gigabyte and Asus, what's the point in putting up with 'good enough' on MSI's?
 
Last edited:
  • Like
Reactions: TJ Hooker

jon96789

Reputable
Aug 17, 2019
414
49
4,740
In case you are wondering, this is my build...

AMD Ryzen 9 3.8 GHz (4.6 GHz) 105-watt 3900X CPU
MSi MPG X570 Gaming Pro Carbon Wi-Fi running BIOS 7B93V11 (AMD PI1.0.0.3a)*
MSi RTX2070 Gaming Z (8 GB) GPU
Corsair LPX 8 GB RAM x 4
Corsair H115i RGB Platinum AIO liquid cooler mounted on top of tower configured as exhaust
Samsung 970 Plus one Terabyte M.2 NVMe SSD
WD six Terabyte Red hard drive x4
LG WH16NS40 Blu-Ray Writer x2
Corsair CX850M PSU
Corsair Obsidian 750D full-size tower with two 140mm fans on front and one 140mm fan on rear configured as intakes
(I could not place the Corsair cooler in the front because the six drive bays did not provide clearance for the rad)

* I did have the latest BIOS 7B93V13 with the AMD PI1.0.0.3abba update installed but the BIOS was buggy and unstable. I think MSi rushed it without fully testing it. The CPU voltages and temps went all over the place. The CPU peaked at 4.5 Ghz (only 0.05 GHz higher than before) but voltage exceeded 1.5 V and the average CPU speed actually dropped in Prime95, from 4.2 GHz to 4.15 GHz. The Power Options settings in the Control Panel no longer worked, the CPU was always running 100% even though I set the maximum power setting to 99%. With the older 7B93V11 I can set the Power Option to Power Saver mode which dropped idle voltage from 1.4875 to 0.95 volts. This dropped the CPU idle temp from 50-55 degrees C to 35-38 degrees. The CPU speed dropped from 3.7 GHz to 2.2 GHz when not under load. When under load, the CPU would peak at a steady 3.725 GHz on ALL cores.

As I mentioned, this PC is used for video rendering and encoding only. Even when using the Power Saver mode and capping the CPU at 99%, my rendering times only increased by a couple of minutes on a 45 minute render time compared to the CPU running at 100% (the app uses the CPU 24-cores more efficiently than the CPU speed). But the CPU and VRM temps never exceeds 75 degrees which is a 20 degree drop and I think better for the CPU life span.
 
Last edited:

jon96789

Reputable
Aug 17, 2019
414
49
4,740
If I were to overclock the CPU, the the temps would go up even higher. Right now, I am running it stock... I would imagine getting a overclocked 3950X would push it even higher than 100 degrees...
 

TJ Hooker

Titan
Ambassador
If I were to overclock the CPU, the the temps would go up even higher. Right now, I am running it stock... I would imagine getting a overclocked 3950X would push it even higher than 100 degrees...
  1. You're running P95, which is generally an unrealistic torture tests that will probably result in higher temps than real world use.
  2. Overclocking is very limited on Ryzen 3K, so I wouldn't expect there to be a huge difference in temps between PBO on and off
  3. An overclocked 16 core CPU is an absolute worst case scenario.
You could also potentially reduce VRM temps just by working with your airflow though, especially with a spot fan pointed at the VRMs. As said above, liquid cooling can sometimes result in higher VRM temps.
Corsair Obsidian 750D full-size tower with two 140mm fans on front and one 140mm fan on rear configured as intakes
Am I understanding this correctly that all three of your case fans are intakes, including the rear fan? Because rear fan should be exhaust.

I'm not saying that MSI's 570 VRMs are great, but it seems like they're still likely 'good enough' for most people (although probably overpriced).
 

jon96789

Reputable
Aug 17, 2019
414
49
4,740
Yes... The three fans are set to intake, the radiator is set to exhaust to create a positive pressure inside the case. The CPU voltage shows VID, not Vcore.

The only way I can drop the voltage is by using the Power Saver mode in control panel. That dropped the voltage to 0.95 volts and the temp down to 36-39 degrees. The ambient temps in the case shows 37 degrees C (I have four thermistors in the case)
 
When most other boards tested between 65-85 degrees, I think 95 degrees would be considered excessive... Some boards were 40 degrees cooler than the MSi. The MSi Godlike and Ace were tested at 65 degrees...
That's bad logic for several reason... First is every VRM design is different and has different operating points including FET temperature, especially under unrealisticly heavy operating load such as an AVX-heavy Prime95 stress test.

Second is FET's commonly have a Tjmax rating of 125C and 95c under extreme (and unrealistic) load such as Prime95 is actually not bad at all.

And last: you can't compare test results unless you precisely duplicate conditions and assumptions made by the other tester. Especially for temperature comparisons where ambient temperature plays a major role: they may be reporting degrees above ambient, for instance, commonly done to normalize for room temperature.
 

brokeBuilder2019

Prominent
Sep 14, 2019
75
15
545

Hi, the mobo review you mentioned is for the Z390 version of the Gaming Pro Carbon. Is there also a review for the one that houses the X570 chipset (i.e. the one the OP has) ? I looked for one, but couldn't find it on Tom's Hardware (reviews on other sites exist).

I'm assuming the differences in the Pro Carbon performance/VRM, etc, across different versions would be significant ? Because of differences in chipset features ?
 

Phaaze88

Titan
Ambassador
Hi, the mobo review you mentioned is for the Z390 version of the Gaming Pro Carbon. Is there also a review for the one that houses the X570 chipset (i.e. the one the OP has) ? I looked for one, but couldn't find it on Tom's Hardware (reviews on other sites exist).

I'm assuming the differences in the Pro Carbon performance/VRM, etc, across different versions would be significant ? Because of differences in chipset features ?
Although I did manage to find this, it's not exactly a review:
View: https://www.youtube.com/watch?v=6BaqXHSEJec&t=2148s
 
  • Like
Reactions: brokeBuilder2019