Tag
A free, open-source book titled 'How to Make Your Model Fast' that provides a systems view of efficient machine learning, from hardware to agents, to help optimize model performance.
The tweet recommends Algorithmica's HPC series section on CPU cache, detailing concepts like latency, memory access patterns, and theoretical latency to help write faster software.
This article provides an introduction to GPU kernels in the context of AI inference and performance engineering, explaining their definition, how they are used, and the advantages of custom kernels for optimization.
Brendan Hansknecht from Modular explains that their high performance numbers stem from treating performance as a full-stack problem, rather than relying on piecemeal components in production.
The article deconstructs the core system architecture of TigerBeetle, a financial ledger database, focusing on performance engineering techniques like static memory allocation and custom zero-copy interfaces to achieve high throughput and predictable latency.
A summary of William Brandon's (performance engineer at Anthropic) GPU programming fundamentals lecture, emphasizing that understanding the streaming multiprocessor (SM) structure of GPU hardware is key to predicting performance, rather than starting solely from the software abstraction of thread blocks/threads.
A detailed thread summarizing the book 'Programming Massively Parallel Processors', focusing on CUDA and GPU programming concepts, optimization techniques, and parallel patterns.
The article introduces core concepts from the book "Systems Performance" regarding latency, throughput, cache hierarchies, etc., and references latency numbers from experts like Jeff Dean, emphasizing the importance of hands-on practice for performance engineering.
Anthropic engineer Tristan Hume discusses the challenges of designing AI-resistant technical take-home tests for hiring performance engineers, detailing how recent Claude models have begun to outperform human candidates.