From CPU to GPU: When Cores Decide
summary
The talk explains when and why to move workloads from CPUs to GPUs, comparing CPU and GPU architectures and performance trade-offs. It covers GPU use cases in modern high-performance systems, practical scenarios where GPUs provide large gains, and guidance on choosing GPU acceleration versus staying on CPU.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:10CPU architecture basics
- 7:51RISC vs CISC processors
- 12:09GPU basics and features
- 16:25GPU memory and compute model
- 20:41Hiding memory latency on GPU
- 25:45GPU clusters
- 28:35When GPU beats CPU
- 31:08GPU-efficient tasks
- 36:58GPU data analytics
- 39:51GPU limitations
- 40:32GPU usage tips
- 41:40Q&A
related talks
59mGPUs - Not Just for Graphics Anymore
50mFrom Magic to Mechanics: Demystifying GPU Programming
37mCo-Location of CPU and GPU Workloads with High Resource Efficiency
01h 06mFoundation of GPU Programming
30mUnderstanding GPUs
33mWhen GPUs Make Python Slower: Patterns and Pitfalls
34mWhat Python Developers Need to Know About Hardware: A Practical Guide...
13mARK: GPU-driven Code Execution for Distributed Deep Learning