AMD, INTEL, NVIDIA, SAMSUNG, SK-HYNIX, MICRON etc. can only be as good as the competition.
We'd not be looking at 16Gbps GDDR6 had Micron not driven the development with GDDR5X proving that we could still scale current GDDR design and can be manufactured reducing costs and increasing frequencies and density.
We had years of 4C/8T processors, AMD came along with a CPU that didn't have to be as good, but one that proved 8C/16T could be put into a $300USD CPU with a sub 100W TDP, i.e we ended up with Coffee-lake and will get 8C/16T from INTEL as well with awesome clocks. Mobile Core i9 is now possible as well, it leverages the same 8 core silicon.
It's competition in how much power can they pack per mm and per watt, which has the indirect effect of boosting performance.
As for reviews:
=============
AMD insists that those seeded with review kits (One 2600X + one 2700X, Two motherboards one from ASUS + one from MSI) sign an agreement which prevents testing of INTEL 8700K etc. from using All core Turbo boost. You may not know it, but that is not INTEL spec, yet board vendors leave it on by default or is the AUTO option sometimes. What that means is, while the Turbo clock for 8700K is 4.7GHz on a single core, board vendors make that 4.7GHz which is obviously way past the rated TDP and presents the CPUs as faster in multi threaded workloads than they are designed to be (this mode exceeds TDP by some margin).
The flip side to this is that, that's how most users experience their Intel CPUs, so to disable this is literally representing the INTEL CPUs in a way that people just don't experience them, especially for the audience that these reviewers write for.
AMD's CPB (Core Performance Boost) and of course as most reviewers didn't notice or go didn't go into detail about is key to all this performance discrepancy. CPB stops working past 37x on the 2700X. That is, you'll not get the single/light thread auto XFR2 OC to 4.35GHz if you set even 38x multi for 3800MHz.
If you're not following - it means AMD has baked in a "All core Turbo Boost" to the only setting that XFR2 can work. Outside of that, you're dealing with a rudimentary frequency management system. Compared to INTEL systems where you can keep single core Turbo even when using manual OC (i.e any multiplier) which is to say your single core/light threaded performance will never suffer because of your manual OC. single core turbo still works even with all CPU cores at 4.7GHz, that one core will jump to 5GHz when needed as on the 8086K, it's 5GHz single core turbo.
That is why AMD has the clause for their approved reviewers and you get the performance numbers you do.
Like against like, 8700K doing 4.7GHz ALL core is quite a bit faster in games than Ryzen 7 2700X regardless of the OC on Ryzen. It's not an IPC advantage but simply clock speed, exactly that which AMD seeks to mitigate via their controlling test conditions. If you test the way they insist, on an 8700K, it means any game that you load will see all cores at 4.2GHz, which is close enough to AMD's all core turbo mode of 3.95 to 4GHz on 2700X. Since we all know clock V clock Ryzen is faster, you end up with results that show INTEL dead on with AMD or sometimes AMD slightly ahead in game tests.