AI ENGVisual Encyclopedia

SCENE 03 / 24 · FOUNDATIONS OF INFERENCE & EXECUTION

TTFT, TPOT and goodput

The three numbers every inference SLO is made of.

TOKENS OUT · 0
TTFTITL × N0FIRST TOKENLAST TOKENE2E LATENCY

Users feel TTFT as responsiveness and TPOT as reading speed. Goodput only counts responses that satisfy BOTH budgets under real load.

TECHNICAL BREAKDOWNModule 1: Foundations of Inference & Execution

TTFT, TPOT, and the goodput discipline

Users feel two distinct clocks: how fast the first word appears (TTFT) and how fast the rest streams (TPOT/ITL). End-to-end latency is their sum over the output length. Throughput counts tokens; goodput counts only responses that met both budgets — the number that actually maps to revenue and SLA credits.

The SLO Split

Product teams promise TTFT < 500ms and TPOT < 50ms far more often than a single E2E number. The split matters because the phases have different bottlenecks and different fixes.

Goodput vs Throughput

A batch of 256 that pushes TPOT to 300ms produces huge throughput and near-zero goodput. Maximizing goodput means sizing batches to the SLO envelope, not to GPU saturation.

Budget Accounting

E2EL = network + queue + tokenize + prefill + N × decode + detokenize. Every middleware hop spends latency the model budget doesn't have.

MATHEMATICAL FORMULATION · GOODPUT RATE
goodput = Σ_requests 1[TTFT_i ≤ B_ttft ∧ TPOT_i ≤ B_tpot] / total_requests

Throughput can rise monotonically with batch size while goodput collapses. The optimal operating point is the largest batch inside the SLO envelope — which is a scheduling problem, not a hardware one.

REAL-WORLD PRODUCTION ENGINEERING
  • Frontier API providers publish TTFT and TPOT separately; internal dashboards should too, per model and per tenant.
  • Latency percentiles (p50/p95/p99) beat means: tail TTFT under load is where user-perceived slowness hides.