Fastware - malloc DESTROYED with pool memory allocator - memory allocator series ( part 4 )

We are exploring patterns and methods which allow us to speed up our software systems, avoid waste and squeeze the best possible performance out of our hardware. This pursuit leads us to analyse large scale design patterns and all the way to small micro optimisations. If you have a passion for fast software and want to learn how to optimise software at any scale, you came to the right place.

If you have questions or suggestions or just need help with a performance problem, email me at [email protected] or message on Twitter @fastware_zukas.

All sources can be found at https://github.com/fastware-zukas

Pull request at https://github.com/fastware-zukas/fastware/pull/1

Useful links:
https://en.wikipedia.org/wiki/Flexible_array_member
https://en.cppreference.com/w/cpp/language/aggregate_initialization
https://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html

00:00 Intro
00:33 Pool allocator concept
03:21 Free-list pool advantages & disadvantages
04:18 Performance diff between pool allocation implementations
06:40 Pool allocator implementation
10:47 Using designated initialisers (https://en.cppreference.com/w/cpp/language/aggregate_initialization)
12:28 Power of 2 block sizes
14:07 Walking thought power of 2 implementation concept
21:40 Continue pool allocator implementation
23:56 The confusion with a flexible array member (https://en.wikipedia.org/wiki/Flexible_array_member)
25:56 Allocate with pool allocator
29:08 Deallocate with pool allocator
33:26 Pool allocator benchmark implementation
34:07 Running performance benchmarks
37:48 malloc DESTROYED and summary