[SOLVED] a question about motherboard thermal performance

Solution
By quality I mean the overall reliability of design, construction, and operation.
....

There are some other things to look at to decipher that kind of 'quality'. For instance: mobo's all have ethernet and audio systems integrated today.

Some ethernet chips are better than others; 2.5Gbit vs 1 Gigabit vs. 100Mb, but also Intel ethernet has a reputation for better service if your system is going into a crowded network.

Some audio chips have much lower noise floor than others and additional features that get you close to broadcast quality audio. That can be important if you're a content creator.

Some motherboards offer more USB type C interfaces and even headers to support front panel type C jacks (if your case has them)...
How hot the VRMs get essentially. At least that's what I think is being measured when they do thermal testing.
Not sure exactly what you mean by 'quality' though...like does it mean it will burn up or something? or just not perform well.

As far as performing well: thermals are VERY important. When the FET's are running extremely hot they're not controlling current well even if within their rated current handling capacity and so that means power delivery to the CPU is unstable. That's when the CPU can crash with a sudden change in processing load.

But if well made with good components that are properly heatsinked, even really HOT the board won't likely 'burn' up, or stop working completely, early in it's life. That's for a couple of reasons: 1) while the components individually have the necessary heat and current handling capacity the circuit design leaves the CPU unstable when the VRM can't adequately control the power delivery and 2) people don't really use their systems "at the limits" except when stress testing. That's why it's become a bit of an ironic joke that people "stress test their computer to death" trying to make a stable overclock.
 
Last edited:
Sep 20, 2020
24
0
10
Not sure exactly what you mean by 'quality' though...like does it mean it will burn up or something? or just not perform well.
By quality I mean the overall reliability of design, construction, and operation. I realize heat, or the relative lack there of, is a very important indicator for this. I was just wondering if there were other metrics that I should be aware of. For instance, are all PCB created equal or is there some kind of grading system? Or how do I know one manufature's PCIe slot is as good as another's?
 

SteveRX4

Notable
BANNED
Sep 29, 2020
1,468
136
890
Most problems with motherboards are from the user choosing one with a poor VRM.
That's why thermal testing has been so well appreciated.
You can also look for motherboard reliability data for specific problems such as pcie slot failure. Most pcie slot failures are due to end user misuse.
 
  • Like
Reactions: charlie_c137
By quality I mean the overall reliability of design, construction, and operation.
....

There are some other things to look at to decipher that kind of 'quality'. For instance: mobo's all have ethernet and audio systems integrated today.

Some ethernet chips are better than others; 2.5Gbit vs 1 Gigabit vs. 100Mb, but also Intel ethernet has a reputation for better service if your system is going into a crowded network.

Some audio chips have much lower noise floor than others and additional features that get you close to broadcast quality audio. That can be important if you're a content creator.

Some motherboards offer more USB type C interfaces and even headers to support front panel type C jacks (if your case has them). Type C requires additional circuitry including power delivery to devices so that can add cost.

PCIe slots can be highly varied: bandwidth in both number of lanes (x1, x4, x16 most common) and even though you may have a 2nd x16 slot on the board it might be wired only for x4. You have to pay close attention to that if you plan on using AIC's.

And an 'armored' PCIe slot for the GPU is a good idea if you're hanging an extremely massive and heavy GPU on it. The problem of pulling the plastic connector off the soldered on contacts did diminish somewhat as GPU's got more efficient and lighter but with triple-wide 3090's coming at us I think it will return. Also: just be sure to unlatch it before trying to tug a GPU out of the socket.

But most of the above can be mitigated if they turn up a problem: a bad ethernet chip can be 'fixed' with a high-quality server grade card installed as an AIC, or use an external DAC for your audio. If running with an extremely massive GPU (3090) use one of the GPU supports brackets that carry it's weight to the top of the PSU. And there are AIC's and powered hubs that offer USB ports in all flavors.

The good thing about those fixes is they're transportable: you can carry them to the next system you buy. That makes those features secondary considerations and lets me focus on VRM power delivery. And memory support.

About memory support. People talk about 'thicker' boards being higher quality without really understanding why. They think it's because their sturdier, but it's because a 10 layer board (vs standard 6 layer or 4 layer) makes it much easier for the designer to route the traces to memory sockets. That means getting extreme high memory overclocks can be easier, or running 4 DIMM's with high clocks becomes possible.

In the end, I don't want to spend a lot of money on LAN, Audio, USB because I'm either content with their shortcomings or use better add-on solutions (a USB hub and TOSLink DAC). I focus on the VRM and, to a lesser extent, the memory support. I don't need a huge amount of memory so 2 DIMM's is always enough.

EDIT: and one last thing about VRM thermal performance. While it may be a good indicator of how far you can push a CPU overclock, IMO it's not a reliable indicator of overall motherboard quality. There are lots of extreme high quality motherboards that would not do well as 'overclockers' because the VRM's would run way to hot (leaving the CPU unstable). They're not INTENDED to run that way, they're intended to run 'on spec'. With top-notch components, optimally rated for the application, they'll last a very long time in service when run on spec. They're not usually going to pop up in any online reviews, though, as they're marketed for commercial and industrial applications, not to consumers.
 
Last edited:
Solution
Sep 20, 2020
24
0
10
Most problems with motherboards are from the user choosing one with a poor VRM.
That's why thermal testing has been so well appreciated.
You can also look for motherboard reliability data for specific problems such as pcie slot failure. Most pcie slot failures are due to end user misuse.
How/where do I find this information?