Session

Machine Mentality: Instructions Per Cycle & Vectors, Why Databases and LLMs Care

In “Machine Mentality: Instructions Per Cycle & Vectors, Why Databases and LLMs Care,” we peel back the layers of modern compute architectures to reveal how low-level encoding and vectorized execution, driven by Instructions Per Cycle (IPC) and Single Instruction, Multiple Data, pipelines, influence both database, data warehouse and large language models (LLM).

What You’ll Learn

** Encoding Strategies & Data Ordering
Explore how Offset, Dictionary, Bit-Packing, Run-Length, and Delta encodings interact with cache lines and prefetchers—and why ordering your data can make or break performance.

** Instruction Throughput & Vectorization
Understand how SIMD‐driven operators leverage wide registers to execute multiple elements per cycle, and why maximizing IPC matters for both query engines and transformer inference.

** Tokenization
Learn how the core of every transformer block consists of a small handful of linear algebra routines, and how using SIMD makes it possible to achieve those tokens-per-second speeds.

Rif Kiamil

I teach and educate about Data, SQL & Google BigQuery

London, United Kingdom

Actions

Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.

Jump to top