This is absolute nonsense.
The page file is a level of memory that
relieves bandwidth elsewhere in the pipeline. I game a lot. I use the page file and never turn it off. Why?
Because it improves system responsiveness, rather than reduces it, and especially when performance is wanted, it helps. I have way too much RAM for gaming, 16GB, never have I seen it at capacity, in fact, it often sits royally below 8GB. Still, the page file is used during gaming, I monitor it in for example Space Engineers, and in that application, turning it off is
harmful to performance of the simulation. Even with many GBs of RAM to spare.
The page file plays a role in scheduling. Its really fantastic you don't see a disadvantage in your 'simple' use case of running a bench and dialing in some frequencies to run it at. But what kind of use case IS that? You're running a single type of load where you tweak memory usage to fit your capacity, on a system built for many concurrent types of loads. You think you're pushing that system. Its laughable, honestly.
Nobody who disables the page file has managed to quantify what performance they've actually gained or lost; I honestly can't even do it for Space Engineers, unless I'd program my path through the simulation for the exact same run. You speak of a huge performance hit if the page file needed to be used - without backing it up. There isn't one. And if there is one, you need more RAM regardless of whether you use it or not
Still though I understand your argument about unnecessary writes on a disk, it stands because if you're not using it, you're 'writing and reading' more directly out of RAM. Sure. But again... can you actually quantify this or is this 99% impression and historical evidence from ages gone by and 1% reality? I think its the latter. I'm using a 2011 Samsung 830 that has always ran a page file and Windows OS and its still in perfect health. That's
eleven years worth of writes and reads in a gaming system.
Things change, in hardware as it progresses, there was also an age where having a filesystem with folder names as brief as you could get them (coded, letter/number, etc.) was considered useful to gain some performance, I'm sure that in the time of kilobytes of data being relevant, this was measurable. Today? You're considered a nutcase for doing so. The same applies here. Old principles are simply no longer relevant.
And those SSDs are definitely not to be used as 'active' disks, but rather for mass storage. They're generally QLC, which also makes them more suited for mass storage rather than high usage-use cases.