Contents Part 3 · vLLM in depth
Part 3 · vLLM in depth
Speed modes in vLLM: optimisation levels, performance modes, CUDA graph choices and attention kernels
What each speed setting changes inside vLLM, measured as start-up time against writing speed on one GPU.
Code zip with this lesson
This lesson is part of the paid Inference Engineering course
The founding price is ₹2,222 or $99$49, one payment. It rises to ₹9,999 or $249 after the founding period, and buying now gives you every part as it is released, at no extra cost, with lifetime access.
Free to study first: Introduction: what this course builds, Before you start, Prefill and decode and The KV cache.
Already bought it? Sign in to read this lesson.