Overclocking for gamers doesn't really do a whole lot for gamers these days.
The turbo boosting on CPUs these days is too good. Check here. At the bottom you'll find the stats for Comet Lake. Let's focus on the 10900K:
https://siliconlottery.com/pages/statistics
Only the top 1% of i9 or i7 chips can hit an all-core turbo of 5.2 GHz. Meanwhile, on a single core, these chips will turbo up to 5.3 GHz. So depending on the game the stock chip will sometimes win the fps showdown.
No, what's more practical are the asymmetrical overclocks. You can see that 100% of chips can be overclocked to +200MHz above the max stock 3-core turbo, or +100Mhz vs. the max stock 6-core turbo. Thanks to TechPowerUp, we can see the stock turbo profile below. Their chip hit 5.0 GHz SSE across 6 threads (3 cores), and 4.9GHz across 12 threads (6 cores). So Silicon Lottery is telling us that 100% of 10900Ks can be manually overclocked to 5.1 GHz on up to 3 cores, or 5.1 GHz on up to 6 cores. The latter would be ideal for gaming, but once again, when you set the frequency like this, your top core won't turbo up to 5.3 GHz, so once again, you're making a trade-off.
However, there is some conflicting information on their chart if you inspect their site more closely. Say you're a lottery winner. You get one of the "golden" samples: Top 1%:
https://siliconlottery.com/collections/cometlake/products/10900kf51g
Those can get up to 5.3 GHz across 3 cores, and up to 5.2 GHz across up to 6 cores.
These raw factors of frequency bear out in actual overclocking reviews. For the Intel 10th Gen, Intel 9th Gen, Ryzen 5000 series, and Ryzen 3000 series, there's virtually nothing gained by overclocking in games. The trade-off is a huge spike in the consumption of power, heat, and associated noise output. This means you need a killer motherboard, CPU cooler, PSU (for a more stable current, not just more power), and finally good ventilation in the case. All of these associated costs mean overclocking only makes sense if you're already on a top-tier system.
3DMark is more practical for confirming an overclock's abstract performance potential with GPU. You can use your GPU software to set a boost frequency above the stock boost. Test where this gets you in 3DMark. Then decide if it's worth the consequences to thermals and noise. Of course, you don't need benchmarking software to do this. You can test it in the games you play with their fps. Verify an average fps for a particular segment of the game, timed, with FRAPS.
There are a few cards like the RX 5700 which have extraordinary overclocking potential if you void warranty by unlocking their power limit (what amounts to a governor set by AMD to limit the value of the card, and create a lower price point). Of course, even with these, how it handles added heat is determined by how effective the cooling solution. So you'll want a 3-fan variant, or best of all, a custom liquid cooling aftermarket design. But the latter are expensive as hell. The opportunity cost of buying one is that you could have bought a more powerful GPU one or two tiers of performance above. So we run into the same issue as we did with CPUs. It really only makes sense to pursue these extra inches if you're on a Scrooge McDuck gaming rig.
After all, when you pay a premium for one of the nicer 3-fan GPUs, these days, what you're really purchasing isn't more performance, or much more performance, anyway, but a video card that will perform about equally while staying much cooler and quieter. Additional features like a no-fan idling mode and dual BIOS are the other perks. Notice no-fan modes have already become a factory feature with the NVIDIA RTX 3000 series and AMD RX 6000 series, and cooling performance with the reference editions is fantastic.
These companies are determined to give the customer everything he needs or wants directly. Long gone are the days of the Sandy Bridge DIY free performance gold rush.