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-cpuI 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.