100x Slower Code due to False Sharing
Explores potential performance degradation of an embarrassingly parallel problem due to false sharing by benchmarking various implementations of a basic kinematic problem. Examples available on Github: https://github.com/KeithJH/kinematics... Chapter Timestamps: 00:00 - What is False Sharing 02:02 - Baseline Implementation 02:33 - Local Atomic Points 03:39 - Global Atomic Points 05:40 - Better Atomic Points 06:49 - Aligned Points 08:08 - Aligned Point 08:26 - Thread Local 08:50 - Array of Arrays 09:14 - Summary Music: Untitled by @keiferjh Examples are similar to those explored in "4x Code Performance with SIMD" ( • 4x Code Performance with SIMD ) and "Increasing code performance with LTO" ( • Increasing code performance with LTO )
going fast is about doing less
I 100%'d the Backyard Nuclear Bomb Building Game
The Moment That Changed Software Development!
Insane Shadow Data Trick in C
Stackmaxxing for a recursion world record
Every Level of Reverse Engineering Explained
Can Too Many Threads Slow Down Your Program? | False Sharing
The 80’s Algorithm to Avoid Race Conditions (and Why It Failed)
Faster than Rust and C++: the PERFECT hash table
The Cost of Concurrency Coordination with Jon Gjengset
Simple Code, High Performance
How Factorio Syncs A Million Objects
Godfather of AI WARNS: We Cannot Stop What's Coming
The fastest sorting algorithm
One Formula That Demystifies 3D Graphics
Fast Inverse Square Root — A Quake III Algorithm
Branchless Programming: Why "If" is Sloowww... and what we can do about it!
UTF-8, Explained Simply
Creator of C++: Bell Labs, Negative Overhead Abstraction, Mistakes | Bjarne Stroustrup
