Why cache beats clock speed
A cache miss is expensive: pulling data from system RAM costs the CPU on the order of 50–100 nanoseconds, while an L3 hit is several times faster. Games have a fairly small working set but touch it constantly within a single frame, so every miss stalls the core while it waits on memory. When that hot data fits inside a large L3, the core has to wait far less often — and that's the gain raw frequency can't match.
Where the difference actually shows
The effect is strongest in CPU-bound scenarios: simulators, strategy games with thousands of units, crowded MMO hubs, and esports titles running at very high frame rates. To check whether you're CPU-bound, run an overlay (MSI Afterburner / RTSS or a built-in counter) and watch GPU usage: if it stays below ~95–97% while FPS stops climbing, the CPU is likely your bottleneck and cache helps here. In pure GPU-bound scenes at 4K, chips with different cache sizes often perform nearly the same.
What to actually do about it
If you're building or upgrading mainly for games, look at chips with extra L3 (the 3D V-Cache lineup) rather than chasing peak gigahertz. On hardware you already own you can't add cache, but you can avoid fighting it: run fast RAM on its rated profile (EXPO on AMD, XMP on Intel), don't let background apps churn and evict game data, and make sure the game runs on the right cores. The tradeoff is real — big-cache chips usually cost more and sometimes clock a touch lower, so for productivity or rendering the win can be smaller than it is in games.
Verdict
For gaming, a large L3 cache is a real, measurable edge rather than marketing — but it pays off mainly where you're CPU-bound and barely shows when the GPU is your limit. If games are the main job, paying up for cache is more often worth it than paying for clock speed; if you game at 4K pinned to the GPU, or you render more than you play, that money is better spent elsewhere.