hard20 pts · first answer countsmodel-servingbatchingperformance
Why batching raises decode throughput
Going from batch 1 to batch 32 in decode barely slows each step but gives about 30× more tokens per second. Why?
Discussion 0
Join free to post in the discussion.
Related questions
- When batch inference fitsMLOps Engineer
- Traffic before the model has loadedMLOps Engineer
- Why export a model to ONNXMLOps Engineer
- Protocol for internal inference callsMLOps Engineer
- Latency and throughputMLOps Engineer
- What TTFT measuresMLOps Engineer
- Gap between streamed tokensMLOps Engineer
- Two phases of LLM generationMLOps Engineer