Efficient Use of Modern x86-Compatible CPUs
summary
Covers techniques for getting the best performance from modern x86-compatible CPUs by addressing issues like caches, branch prediction, instruction-level parallelism and memory ordering. Discusses practical approaches such as low-level profiling, cache-friendly data layouts, lock-free and concurrency-aware algorithms, and CPU-specific optimizations to improve throughput in high-load systems.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 3:59Local multithreading
- 11:39Vector processor commands
- 20:18Data alignment and packing
- 26:40Writing assembly code
- 33:17ISPC program acceleration
- 37:14CPU caches and data handling
- 40:40Q&A
related talks
44mOptimizing Software for Modern CPUs and Linux
54mx86 Internals for Fun & Profit
01h 05mMegahertz, Gigahertz, Registers and Instructions: How does a CPU actually work?
48mPerformance and Cost: How Hardware Influences Our Programs
01h 11mPerformance engineering - being friendly to your hardware
01h 02mMegahertz, Gigahertz, Registers and Instructions: How does a CPU actually work?
51mLinux Systems Performance - Architecture
38mPeculiarities of modern hardware