• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

What are you looking for in a Motherboard Review?

That's something you can work out once you know the boards maximum VRM wattages anwyay


If a board can only handle 200W without throttling, you already have the answer to what CPU's are supported. Instead of testing a dozen CPU's, you can test one high end model and discover the wattage limits, and then look up the wattages of the various CPU's. Same information, far less work.
I don't care about watts, because they don't tell you what frequency CPU will be stable at, Voltage however - does.
At least to me "this cheap VRM can cope with 7950X up to 1.05V, and 7700X up to 1.2V", sounds more helpfull than "this cheap VRM will limit everything to 200W". You don't need to test dozen of CPUs, just test each die configuration once.
I simply think from stable performance perspective, vs. simple approach of "I don't care about performance, just limit to whatever, whenever needed".
 
That's something you can work out once you know the boards maximum VRM wattages anwyay


If a board can only handle 200W without throttling, you already have the answer to what CPU's are supported. Instead of testing a dozen CPU's, you can test one high end model and discover the wattage limits, and then look up the wattages of the various CPU's. Same information, far less work.


Why not? Motherboards don't add any performance - the only time a motherboard can be a performance detriment is this one situation of VRM's that cant handle stock


An A320 runs like an x570 with no performance change
(Random screencaps from youtube, first result when searching)
View attachment 261453
View attachment 261454


By that metric, it's absolutely worth knowing if a motherboard is going to hamper performance - if they say they support high end CPU's on a budget board, they better deliver
Don't get me wrong, I'm not looking down on budget chipsets. :) I used to have an Asus TUF A520M which is a brilliant board (all of the latest Asus TUF boards are good, in fact).

My point is the same as yours: if one wants to run a high-end CPU, then one had better make sure the VRM delivers. Bottom of the barrel boards tend to be quite bad in that regard.

This is a classic video on the topic despite the clickbait title:

 
Adding to VRM testing point :
In lower end or problematic motherboards reviews, I would love to see how much negative offset (-0.1V or -0.2V) on Vcore can help them cope with higher CPU models (if they are supported). Obviously with decresed max. turbo ratios by 100 - 400MHz when they unstable, or for "die quality lottery" margin to make everyone happy.

Last month I wanted to buy cheapest B660 full ATX MB, but there were simply no tests/reviews of them anywhere :(
I eventually picked Gigabyte's B660 DS3H DDR4 "open box" with discount.

As a side note, on Intel, when using default Adaptive voltages one should preferably not use offsets to tweak them. Adaptive voltage control works best with a different tuning knob called "AC Loadline" which increases voltage proportionally to current (in amps) into the CPU.
 
From my experience low end MBs can do gaming just fine with any CPU. But if you do 3D Rending / Heavy Encoding, the vcore drops so low it auto restarts.

Overclocking is generally out of the question because it is either missing the settings or the MB cant sustain the requested voltage.

I really think AMD and Intel should be stricter on the rules for allowing top end cpu in budget MBs. It just makes them look bad when it doesn't work correctly. Like for example, the MSI B660M Mortar I reviewed had no overclocking features, which is fine (budget MB), but it couldn't handle the full load of a 12900K (stock).
 
Your saying that because most use one nvme

2 here! One PCIe 3.0 and one PCIe 4.0.... :D

Got 5 M.2 slots total, so when I need more storage in the near future , it will be NVMe SSD's..
 
I really think AMD and Intel should be stricter on the rules for allowing top end cpu in budget MBs. It just makes them look bad when it doesn't work correctly. Like for example, the MSI B660M Mortar I reviewed had no overclocking features, which is fine (budget MB), but it couldn't handle the full load of a 12900K (stock).

Not even by increasing what MSI calls "CPU Lite Load"? The "AC Loadline" voltage correction I mentioned above is behind that (setting control to "Advanced" makes it clearer).

Intel recommends by default to set it to the same impedance of the VRM, which should be 1.1 mOhm for 8-cores ADL Intel CPUs. A value this high is conservative and should prevent voltages from decreasing too much.

But if the motherboard has silly defaults and uses a higher VRM impedance (e.g. the 1.7 mOhm value intended for 6-core CPUs), then voltage will drop too much under load. VRM impedance would be otherwise adjusted by the LLC setting.

This is another possible reason why default motherboard settings need to be very carefully checked as they might not be actually "optimal" at all.

Last year I had a Gigabyte B560 motherboard which couldn't apparently handle full load with a 11700K at default settings, resulting in crashes. All it needed was a higher LLC setting and a higher AC Loadline than default.

1662998238812.png


The image above also shows inappropriate defaults: Power limits at 4096W and Current Limit essentially unlimited (512A). Excessively high limits, in particular current, can also cause instability.

The 12600K is supposed to have a 175A current limit and the 12900K a 280A current limit, and unlike power limits, these are not intended to be exceeded.
 
Last edited:
Not even by increasing what MSI calls "CPU Lite Load"? The "AC Loadline" voltage correction I mentioned above is behind that (setting control to "Advanced" makes it clearer).
Hmm, its been a bit but if I remember correctly the solution to the problem was to actually max out the LLC and PWM. Maybe future BIOS updates resolved this, but Auto which 99% of users will leave it on could not handle a stock 12900K under full load. Works great with a 12600K though and that's what its being used with now.

Because I use a waterblock its on 4096W by default (auto settings).
 
I agree that most end-users will use default settings, but that's in my opinion where reviewers could help. People generally assume that defaults are optimal or stable, but it's often not the case. Then, like bad hardware is often called out, bad BIOS settings should be called out too.



On a related note, on modern MSI boards, CPU Lite Load should become like this when set to "Advanced", exposing Loadline values more directly in hundredths of mOhm.

1663002015820.png


AC Loadline directly affects voltages (when using Adaptive CPU voltages), DC Loadline affects only the reported values. When DC Loadline has been set to the correct value (which depends on the LLC setting), CPU VID will match Vcore under all-core loads. This is also a way for finding out the impedance of the LLC setting used.

Maxing out LLC is generally not advised due to potentially damaging/destabilizing voltage over/undershoot during load transients that can only be observed with an oscilloscope, and decreased VRM efficiency.

The VRM PWM setting may or may not have negative effects; on my MSI Z690 A-Pro, setting it to the maximum of 677 kHz makes no difference whatsoever to temperatures or total system power consumption.
 
Last edited:
I agree that most end-users will use default settings, but that's in my opinion where reviewers could help. People generally assume that defaults are optimal or stable, but it's often not the case. Then, like bad hardware is often called out, bad BIOS settings should be called out too.
If an officially supported CPU doesn't work in a motherboard with default/auto settings, then it's a rubbish motherboard. The purpose behind custom settings is for enthusiasts to tweak stuff, not to make things work at all.
 
If an officially supported CPU doesn't work in a motherboard with default/auto settings, then it's a rubbish motherboard. The purpose behind custom settings is for enthusiasts to tweak stuff, not to make things work at all.
I agree. Either the MB needs its auto settings tweaked or remove support for the CPUs that won't run on auto. I test everything on Auto, unless I can't complete the testing lol. That would have been a very short review. "Sorry no testing was done because I couldn't be bothered to manual change the settings".
 
Somethings that most reviews don't have and are nice to know about a motherboard.

1. Boot Sequence
- From Cold Boot to BIOS/UEFI settings
- Cold Boot to OS
2. Onboard Audio in comparison to best in class add on audio cards/adapter
3. Display outs capability
4. USB Port functionality (How many share the bandwidth/which ones have the full bandwidth without sharing with other ports)

I will add more...
 
If an officially supported CPU doesn't work in a motherboard with default/auto settings, then it's a rubbish motherboard. The purpose behind custom settings is for enthusiasts to tweak stuff, not to make things work at all.

In my opinion motherboard defaults should strictly set Intel (or AMD)-recommended limits and settings, but it's almost never the case. Furthermore, in some cases motherboards don't even make it simple to configure "true" default settings.

For instance, at first boot, recent MSI Intel motherboards present a choice for setting power and current limits depending on the installed CPU cooler, which probably @ir_cow also observed for the MSI B660 review. In my case, the base/minimum choice ("Boxed cooler") set PL1=PL2=241W. But my 12700K is supposed to have PL1=125W, PL2=190W, Tau=56s by default...

Also on the Gigabyte Aorus B560 Pro AX I previously had, motherboard defaults were "no power limits", which coupled to insufficient default voltage/excessively drooping voltage, made for unstable operation at high loads. It had an option for applying Intel-recommended limits, but these were partially incorrect (with PL2=PL1*1.25, which is the hardware default, not the recommended value) and under a cryptic name ("Turbo Power Limits: POR").

In general, my recent experience with default settings with middle-range motherboards has not been good. It's not that the products themselves were defective or lacking in the hardware, but less experienced users may easily think so if they encounter similar issues. Many people also often complain about the high power consumption of modern systems. Default motherboard settings are to blame for this.
 
In my opinion motherboard defaults should strictly set Intel (or AMD)-recommended limits and settings, but it's almost never the case. Furthermore, in some cases motherboards don't even make it simple to configure "true" default settings.

For instance, at first boot, recent MSI Intel motherboards present a choice for setting power and current limits depending on the installed CPU cooler, which probably @ir_cow also observed for the MSI B660 review. In my case, the base/minimum choice ("Boxed cooler") set PL1=PL2=241W. But my 12700K is supposed to have PL1=125W, PL2=190W, Tau=56s by default...

Also on the Gigabyte Aorus B560 Pro AX I previously had, motherboard defaults were "no power limits", which coupled to insufficient default voltage/excessively drooping voltage, made for unstable operation at high loads. It had an option for applying Intel-recommended limits, but these were partially incorrect (with PL2=PL1*1.25, which is the hardware default, not the recommended value) and under a cryptic name ("Turbo Power Limits: POR").

In general, my recent experience with default settings with middle-range motherboards has not been good. It's not that the products themselves were defective or lacking in the hardware, but less experienced users may easily think so if they encounter similar issues. Many people also often complain about the high power consumption of modern systems. Default motherboard settings are to blame for this.
Well that's why they're all labelled Auto not default in the bios, so the AIB can mildly OC your part and claim they're boards the best.

Some don't even have a default option, just auto(NOT default) , Xmp or manual.

This is a generalized statement that's true of every recent board I have seen.

And I agree, it's not good and Should be stopped.
 
In my opinion motherboard defaults should strictly set Intel (or AMD)-recommended limits and settings, but it's almost never the case. Furthermore, in some cases motherboards don't even make it simple to configure "true" default settings.

For instance, at first boot, recent MSI Intel motherboards present a choice for setting power and current limits depending on the installed CPU cooler, which probably @ir_cow also observed for the MSI B660 review. In my case, the base/minimum choice ("Boxed cooler") set PL1=PL2=241W. But my 12700K is supposed to have PL1=125W, PL2=190W, Tau=56s by default...

Also on the Gigabyte Aorus B560 Pro AX I previously had, motherboard defaults were "no power limits", which coupled to insufficient default voltage/excessively drooping voltage, made for unstable operation at high loads. It had an option for applying Intel-recommended limits, but these were partially incorrect (with PL2=PL1*1.25, which is the hardware default, not the recommended value) and under a cryptic name ("Turbo Power Limits: POR").

In general, my recent experience with default settings with middle-range motherboards has not been good. It's not that the products themselves were defective or lacking in the hardware, but less experienced users may easily think so if they encounter similar issues. Many people also often complain about the high power consumption of modern systems. Default motherboard settings are to blame for this.
I tend to avoid Gigabyte products, to be honest. Their BIOSes are cryptic, and the general quality is not there where it should be (at least it wasn't the last time I had something from them).

MSI is generally OK quality-wise, but I find them lacking on the BIOS front (cooler choice for TDP? Come on)...

ASUS is the best all around, in my opinion. Their quality is good, the BIOS has all the right labels and defaults to Intel / AMD recommendations (at least the ones I've had so far do). I wouldn't have anything else in this day and age - and that's not brand loyalism, just it's only ASUS that offers the quality and features that all mid-range and high-end motherboards should.
 
I don't care about watts, because they don't tell you what frequency CPU will be stable at, Voltage however - does.
At least to me "this cheap VRM can cope with 7950X up to 1.05V, and 7700X up to 1.2V", sounds more helpfull than "this cheap VRM will limit everything to 200W". You don't need to test dozen of CPUs, just test each die configuration once.
I simply think from stable performance perspective, vs. simple approach of "I don't care about performance, just limit to whatever, whenever needed".
wattage is something that applies to any CPU used in the board.

voltage is worthless since no two boards ever agree on them, with all the Vdroop and LLC settings out there and different autos between boards, let alone the differences that pop up between VID and voltages... it's utterly meaningless.

If you know a board can do 200W on AM4, you know it can handle literally any CPU at stock since none pass 142W - and that you can overclock a 5900x or 5950x to that 200W limit.

Same goes for intel, if you know the CPU's need 250W to sustain 5.2GHz, why care what voltage shows up? It's an average of an estimated software reading, and every single CPU will be different anway
 
I tend to avoid Gigabyte products, to be honest. Their BIOSes are cryptic, and the general quality is not there where it should be (at least it wasn't the last time I had something from them).

MSI is generally OK quality-wise, but I find them lacking on the BIOS front (cooler choice for TDP? Come on)...

ASUS is the best all around, in my opinion. Their quality is good, the BIOS has all the right labels and defaults to Intel / AMD recommendations (at least the ones I've had so far do). I wouldn't have anything else in this day and age - and that's not brand loyalism, just it's only ASUS that offers the quality and features that all mid-range and high-end motherboards should.
I disagree on Asus following CPU specs on board's, they use Auto as a Smoke screen of BS, not all have a default option in the list even and they love upping the CPU voltage.

Though if you know this they are, to me still the best board maker.
 
wattage is something that applies to any CPU used in the board.

voltage is worthless since no two boards ever agree on them, with all the Vdroop and LLC settings out there and different autos between boards, let alone the differences that pop up between VID and voltages... it's utterly meaningless.

If you know a board can do 200W on AM4, you know it can handle literally any CPU at stock since none pass 142W - and that you can overclock a 5900x or 5950x to that 200W limit.

Same goes for intel, if you know the CPU's need 250W to sustain 5.2GHz, why care what voltage shows up? It's an average of an estimated software reading, and every single CPU will be different anway
Since CPU Watts are made from Volts supplied by MB's VRM, then that means when CPU Voltage isn't stable, watts also can't be stable.

Q : How do you know you can sustain 250W at 5.2GHz accross different boards, when voltage differences between boards are too big for it to be meaningfull ?
 
I disagree on Asus following CPU specs on board's, they use Auto as a Smoke screen of BS, not all have a default option in the list even and they love upping the CPU voltage.

Though if you know this they are, to me still the best board maker.
My TUF B560M Wifi uses Intel-defined power limits as Auto (65 W PL1 and 228 W PL2 with 28 s Tau on my 11700). I don't know about voltage, but it is so dynamic nowadays that I find it best not to touch it.
 
My TUF B560M Wifi uses Intel-defined power limits as Auto (65 W PL1 and 228 W PL2 with 28 s Tau on my 11700). I don't know about voltage, but it is so dynamic nowadays that I find it best not to touch it.

Dynamic voltage control on Intel is not as sophisticated as it seems, unfortunately.

As for recent ASUS motherboards, they have an easily-accessible "SVID behavior" configuration option. Behnd the scenes, this adjusts the "AC Loadline" voltage correction I mentioned earlier according to various presets. As far as I am aware of, "Best-Case Scenario" sets 0.01 mOhm (no voltage correction; the CPU can end up being unstable), "Intel's Fail-Safe" sets 1.1 mOhm (assuming an 8 p-core CPU and default LLC; can end up overvolting the CPU especially if the LLC setting has been changed). See attached screenshots.

In theory, a value of 1.1 mOhm implies that the preprogrammed base voltage-frequency curve is corrected up by 1.1 mV for each A of current into the CPU. So, if the VRM had the same impedance (that is, caused voltage to drop with current by the same amount), voltage would remain constant for a certain frequency no matter the load. In practice, things might not work so well.

That value should also be able to be more directly configured in advanced options elsewhere.
 

Attachments

  • 1663326379072.png
    1663326379072.png
    463.3 KB · Views: 179
  • 1663327006493.png
    1663327006493.png
    252.8 KB · Views: 117
Well,seeing as I was currently shopping for a new board. Fan headers and placement. I need front headers for my intake fans and my pump.
if you look at my current specs my board has 3 headers on the front of the board it had a total of 6. I almost bought a used Aorus Elite that had ZERO on the front and a total of 4 thats including the 2 CPU headers.
i had to totally step into unknown territory and settled on an MSI board because X570 boards are getting hard to find and I’m just desperate to get back up and running.Thankfully this new board will totally meet my needs and pretty Sh1t Hot board to boot.
 
Since CPU Watts are made from Volts supplied by MB's VRM, then that means when CPU Voltage isn't stable, watts also can't be stable.

Q : How do you know you can sustain 250W at 5.2GHz accross different boards, when voltage differences between boards are too big for it to be meaningfull ?
Because stock voltages are always excessive for this exact reason


You're truly missing the point of how useless voltages comparisons would be, not a single person would be able to benefit from that time intensive testing, except the person who owns that EXACT CPU - not the same model or series, but THAT one single CPU.


Reviewer would need a golden sample, cherry picked to get any good results... and they'd be useless. Or they'd get a dud, and again... useless.
 
Since CPU Watts are made from Volts supplied by MB's VRM, then that means when CPU Voltage isn't stable, watts also can't be stable.

Q : How do you know you can sustain 250W at 5.2GHz accross different boards, when voltage differences between boards are too big for it to be meaningfull ?
A: As @Mussels pointed out you can't compare with cpu voltage since no one CPU is the same. That extends to cpu freq. A work around to that is to set the voltage to a static value. Raise the freq until the desired wattage. Combine both togther and you have a VRM test which encompass all types of CPU loads. A "worst case" if you will for normal CPU use.
 
Last edited:
Back
Top