What's the difference between vTaskDelay and vTaskDelayUntil, and when does it matter?

vTaskDelay blocks for a duration measured relative to the moment it's called, so the actual period between iterations is your work time plus the delay, and since work time varies, the period drifts and accumulates jitter. vTaskDelayUntil (or its newer xTaskDelayUntil form) instead blocks until an absolute wake time that you advance by a fixed increment each iteration, so the task wakes on a precise, fixed cadence regardless of how long the work took (as long as it fits within the period). It matters for anything that must run at a steady rate, a control loop sampling a sensor and updating an actuator every 10 ms, a periodic protocol heartbeat, where drift or jitter degrades the algorithm or violates timing. The mental model: vTaskDelay says "sleep for 10 ms," which compounds with your runtime; vTaskDelayUntil says "wake every 10 ms," anchoring to a schedule. The caveat is that if the work ever exceeds the period, vTaskDelayUntil can't catch up and you've simply overrun, which is itself useful information that the task is over budget. For one-off or non-periodic waits, vTaskDelay is fine; for fixed-rate periodic tasks, use vTaskDelayUntil.