Name: You asked for it, so here it is. Visualizing CPU cache speeds relative to RAM. Cache optimization is important too!
Uploaded: 2024-10-18T16:13:20.000Z
Duration: PT15.333S
Channel: Ben Dicken
Description: Ben Dicken shorts video about You asked for it, so here it is. Visualizing CPU cache speeds relative to RAM. Cache optimization is important too!

You asked for it, so here it is. Visualizing... show more

Ben Dicken

1,474,564 Aufrufe • vor 1 Jahr

Some of the biggest news today is Cloudflare Worker's... show more

Jilles Soeters

17,920 Aufrufe • vor 1 Monat

Submitted Easy Vertex Anim for #UE5.8 too. You can... show more

Inu Games

10,373 Aufrufe • vor 2 Monaten

I don't think anyone realizes yet how powerful Convex... show more

Rhys

27,547 Aufrufe • vor 1 Jahr

WanGP's latest MiniMax H3 update is huge: - Way... show more

cocktail peanut

58,539 Aufrufe • vor 25 Tagen

A lot of people asked me to post FOOTage... show more

Geralt of Goonia

261,127 Aufrufe • vor 4 Monaten

Want the latest, flashiest skins? Lucky Cache is here... show more

Call of Odin's Chosen

121,198 Aufrufe • vor 8 Monaten

Day 11/90 of Inference Engineering How does vLLM work... and how is it used in production? Before we discuss how vLLM works internally, it helps to understand what vLLM is. At a high level, vLLM is an inference engine that is designed to serve LLMs to thousands of concurrent users efficiently while managing scarce compute and memory. The goal for vLLM is to maximize throughput and minimize latency; optimizing for the best inference economics and experience for end users. With every request from the end user, it eventually ends up in the engine core, gets scheduled alongside other requests from other concurrent users, executes on the GPU, and updates the KV cache with the new key and value vectors, and streams the tokens back to the user. The Scheduler decides what requests should execute next while continuously batching requests together to maximize GPU utilization. Continuous batching is an inference optimization that allows new requests to join a running batch as other requests finish generating tokens. This helps with keeping the GPU utilization high instead of letting it sit idle waiting for an entire batch to complete generating. After the scheduler dispatches the selected batch to the Model Executor, the Model Executor prepares the tensors and metadata required for inference, retrieves each request’s block table from KV Cache Manager, launches the optimized transformer forward pass on the GPU, computes the logits, updates the KV cache with the new key and value vectors, and finally returns the results for sampling and streaming. The KV Cache Manager uses the PagedAttention memory layout to allocate fixed-size cache blocks on demand and maintains a Free Block Queue on the CPU that tracks which blocks in the GPU’s Paged KV Cache are currently free. When a request needs additional KV cache space, the KV Cache manager takes a free block from the queue and assigns it to that request, thus avoiding an expensive search through GPU memory for available cache blocks. All of these components form the core of vLLM’s inference engine. The Scheduler determines what requests are executed, the Model Executor determines how those requests are executed, the KV Cache Manager determines where each request’s KV cache lives using the PagedAttention Memory Layout. This architecture enables vLLM to serve thousands of concurrent requests with high throughput, low latency, and efficient GPU memory utilization. Heres a little animation that visualizes everything! - I've also completed the forward pass for my mnist.c project. I had a nice chat with shrey birmiwal, such a knowledgeable guy. Excited to learn more about vLLM and implement a tiny-vLLM one day.show more