Keyboard Matrix Scanning, N-Key Rollover (NKRO) and Ghosting Elimination: Competitive Benchmark Analysis and Hårdvara Interactions

[Gabriel Reed]

I’ve been stress-testing three different matrix architectures over the last few weeks—cherry MX Brown with full NKRO, a budget 60% with 6KRO, and a hot-swappable gasket with a STM32 controller—and the frametime variance between them is genuinely measurable in-game, not just on paper. Before I run a bigger benchmark, I want to hear from people who’ve actually dug into the scanning side of things. At what point does a lower rollover count actually start costing you inputs during a rapid ability sequence, and do you think the bottleneck is the diode matrix or the microcontroller’s polling loop? I’m especially curious about whether the firmware debounce delay is masking ghosting issues that would otherwise show up in raw scanning data.

Also worth discussing: how much does matrix scanning frequency genuinely matter versus just having reliable NKRO for competitive titles? I’ve seen claims that 250Hz polling versus 1000Hz is negligible for keystroke registration, but I want empirical data. Has anyone done controlled tests correlating input latency with frametime consistency, or is most of the variance just noise from driver overhead? And for the hardware scaling question—when you move to a faster controller, where does the real bottleneck actually shift?

"Wooting 60HE (Tachyon mode enabled, 0.1mm rapid trigger) + Lamzu Maya 4K"

[Author]

Posts