Which scope for this i need ?

[Mr. Scram]

For what you get, low end commercial MOBOs are cheap at around $50.  OTOH, a high end gaming MOBO might be fairly pricey.  The last one I bought was about $270 or a little over 10% of the project cost.  Speed costs money!  How fast can you afford to go?

You can buy an equivalent performing model for half that. Most expensive motherboards are loaded with all sorts of excess featues. Unless you're talking about the more exotic platforms, in which case you actually have massively wide buses and such. Those simply cost copper and therefore money.

[hans]

High-end motherboards have:

- Top tier chipsets instead of business models.
- More ports, usually added with additional chipsets because the "vanilla experience" (i.e. what is offered by the chipset) didn't include these. That can be extra USB3.0, Thunderbolt, Wifi, more LAN ports, more SATA ports, etcetera.
- More comprehensive PCI-e switching arrangements. Typically some PCI-e switches are used to reconfigure slots depending on occupancy. I.e. 1x16 or 2x8 switching.
- More beefy power supply, aimed at overclocking
- (More) High-end audio solutions
- "Better validation", although it's hard quantify this really. Some configurations may break a system, some don't.

If you're just looking at stock settings, then there is absolutely no performance gain in a high-end board. Value of these boards is defined differently, like I just explained, in terms of extra connectivity, 'stability', overclocking potential and some part also branding (does it have RGB? ).

[rstofer]

For what you get, low end commercial MOBOs are cheap at around $50.  OTOH, a high end gaming MOBO might be fairly pricey.  The last one I bought was about $270 or a little over 10% of the project cost.  Speed costs money!  How fast can you afford to go?

You can buy an equivalent performing model for half that. Most expensive motherboards are loaded with all sorts of excess featues. Unless you're talking about the more exotic platforms, in which case you actually have massively wide buses and such. Those simply cost copper and therefore money.

It's easy to say a MOBO has matching or superior performance but there is no way for a user to prove it short of buying a bunch of boards and trying them with their particular applications.  The only solution I have found is to drive a stake in the ground and call it good.

Furthermore, what is there to test?  I don't buy gaming MOBOs for gaming, I don't even play games.  All I want is speed when running Vivado.  But Vivado won't use more than 8 threads so there is no point in going after MOBOs for dual or quad 16 core 32 thread Xeon chips.  In any event, gaming is more a function of the video card than the MOBO.

Support for higher speed memory seems important, especially for programs that use a lot of RAM.

What I don't understand is why anybody wants to waste electricity on LED theme lighting on the internals.  My PC isn't a toy!  I just want enough speed to synthesize a circuit before I die of old age.

[Mr. Scram]

It's easy to say a MOBO has matching or superior performance but there is no way for a user to prove it short of buying a bunch of boards and trying them with their particular applications.  The only solution I have found is to drive a stake in the ground and call it good.

Furthermore, what is there to test?  I don't buy gaming MOBOs for gaming, I don't even play games.  All I want is speed when running Vivado.  But Vivado won't use more than 8 threads so there is no point in going after MOBOs for dual or quad 16 core 32 thread Xeon chips.  In any event, gaming is more a function of the video card than the MOBO.

Support for higher speed memory seems important, especially for programs that use a lot of RAM.

What I don't understand is why anybody wants to waste electricity on LED theme lighting on the internals.  My PC isn't a toy!  I just want enough speed to synthesize a circuit before I die of old age.

That's what reviews are for, and sometimes some system builders also do thorough testing. Look at a few, find out what your sweet spot is and you're done. I'm not sure about Vivado, but I've seen some extensive testing being done in regards to several engineering software suites.

Why anyone would RGB, or a colour themed motherboard? The same reason why some people kit out their cars. To some, a car is an object that aids travel from a to b. To others, it's a hobby and an expression of what they perceive themselves to be.

[hans]

Benchmark of components: look at the right reviews.
E.g. https://uk.hardware.info/reviews/7652/9/18-x299-motherboards-review-battle-for-first-benchmarks-cpu

I don't think you'll often find performance comparisons between a whole range of boards for a particular platform. Some of these benchmarks are synthetics, some are real-world like x264 encoding. What is applicable? Depends on your application, and if you want to make any judgements yourself, you better get to know it.
As you can see, practically any board from "bottom of the range" up to "top end" performs the same. There are some outliers, but it almost goes without saying that these boards have OC profiles applied "out of the box", mostly being enthusiast grade level boards. This was especially a huge shitstorm when the i7 8700K coffee lake CPU's landed, and some reviewers with non-Asus boards got a Cinebench score in the 1300 region, and some others with Asus boards got 1500+, while a few reviewers with Asus boards got 1300 points again.
Reason: there was a Asus-specific "enhanced multi-core experience" setting enabled by default, which allowed for higher turbo clocks for longer, as long as cooling permits. Some reviewers turned this off, some others didn't, some others didn't have Asus boards.. so they didn't have this feature.
This also somewhat concludes that benchmarking is hard, it's easy to get the results you want, and not the ones you should get.

But in most cases, stock settings are applied: stock clocks, stock memory speeds, all the advertised speeds by Intel ARK. You can buy 4000MHz RAM, but it will probably run at the "validated" 2333 or 2666MHz. Unless you manually go in there and load a XMP profile or put it up higher. Similarly, you could go manually into your BIOS and try to OC your system to gain more performance. But honestly anyone that is targetting a workstation application would leave this kind of stuff out of their damn mind.

If you need to figure out what combination of parts works best for your application, then motherboard reviews can be pretty useless. They are basically a non-factor, unless you need a very specific I/O benchmark, like USB3.0 transfer speeds..
Look up a CPU review and figure out how your application scales with memory bandwidth, CPU single-core and multi-core performance, and how well these demands fit in the offerings of AMD and Intel.

Puget systems also performs a lot of these benchmarking, alot of it is available to the open public in their publications. Unfortunately, a lot of focus is on media creation though, as it tends to have alot of benefit from GPU acceleration if done right.

[Mr. Scram]

Puget Systems does some great work, separating facts from nonsense. It's the kind of testing that's very impractical and expensive to do yourself, but which is invaluable if you want to tailor a system to your needs without making wild guesstimations.

[nctnico]

When it comes to PC hardware I want stability and reliability before performance. After all: what good is a computer if it can produce the wrong result or no result at all due to a crash/software malfunction?

[Mr. Scram]

When it comes to PC hardware I want stability and reliability before performance. After all: what good is a computer if it can produce the wrong result or no result at all due to a crash/software malfunction?

On the other hand, having all the stability in the world isn't great when you don't have usable performance. Performance is as important to get the job done, but preferably not at the cost of stability.