Thirty against sixty

The Trade

At sixty frames a second you have roughly sixteen and a half milliseconds to simulate, render, and push a frame to screen. Drop to thirty and that doubles to thirty-three. On a fixed platform, where the hardware will not change beneath you, this is the most reliable way to buy room — not just a little, but a lot.

The decision is made early, because it touches almost everything downstream. Animation systems assume a tick rate. Physics runs at a fixed step. Audio callbacks are timed. AI budgets are set against how often they fire. A target chosen at the start of production and reversed late is not a free swap — it is a rebalance across most of the codebase.

Sixty frames is the right answer for anything where input latency is a competitive concern: fighting games, shooters, anything where a player's read of the screen leads directly to a motor response. The penalty for dropping below sixty in those genres is felt in the hands before it registers in the head. Thirty is defensible — and often invisible — in slower-paced games: open-world exploration, strategy, narrative work, turn-based systems where the game is not reacting to a reflex.

The middle answer, forty, has become more available as displays supporting variable refresh rates become common on consoles. It does not divide cleanly against a sixty-hertz screen, which historically made it a non-starter, but on a screen that can run at 120 Hz it sits as a clean third tick: more fluid than thirty, less demanding than sixty. Some recent console releases have shipped with a menu option that targets forty specifically for this reason.

The failure mode is indecision — shipping a performance mode and a quality mode because the team could not agree which mattered more, then discovering that neither mode was playtested seriously because two targets meant half the coverage on each. Both modes launching with their own distinct bugs is the common result.

Picking a number and holding it is itself a production decision. The target frame rate is not a graphics setting. It is a constraint that runs through profiling, physics, AI, and audio in equal measure, and deferring it is the same as deferring all of those conversations at once.

A target chosen at the start of production and reversed late is not a free swap — it is a rebalance across most of the codebase.

Every figure in this entry is one worked example on one platform, shown because a partition makes the trade-off legible. It is not a measurement of any particular production.