A living knowledge base of C++ internals, low-level systems, and performance engineering.
Every post published on LinkedIn has corresponding code here. The repo grows daily. Clone any folder, build it, and run the benchmarks yourself.
Status: updated daily.
Each LinkedIn post covers one specific concept: a CPU quirk, a compiler behavior, a memory model subtlety, or a performance trap. The code here is the runnable, benchmarked version of whatever was posted.
| Folder | Topics |
|---|---|
| language/ | Core language mechanics, ODR, lifetime, value categories, copy/move semantics |
| compiler/ | Codegen, optimization passes, inlining, constant folding, reading assembly |
| cpu/ | Caches, branch prediction, pipeline stalls, false sharing, prefetching |
| memory/ | Allocators, alignment, NUMA, memory ordering, cache-line layout |
| intrinsics/ | SIMD, SSE/AVX/AVX-512, manual vectorization, throughput vs latency |
| stl/ | STL internals, iterator invalidation, allocator-aware containers, SBO |
| templates/ | Template metaprogramming, SFINAE, concepts, CRTP, compile-time cost |
| performance/ | Profiling, perf, flamegraphs, PGO, LTO, hot/cold splitting |
| low-latency/ | Lock-free, wait-free, RDTSC, kernel bypass, CPU pinning, OS jitter |
| concurrency/ | Atomics, acquire-release, sequential consistency, data race analysis |
Each topic is a subdirectory under its domain, named in lowercase with hyphens:
/<domain>/<topic>/
README.md
benchmark.cpp
main.cpp
benchmark.cpp holds the example and timing.
main.cpp is added when the concept needs a standalone demo separate from the benchmark.
Each topic's README covers the concept, what the code shows, the compile and run commands, and the actual output from running it.
Browse by domain using the table above. New folders are added as posts go out, so a domain may be empty until its first topic lands. Topics post daily.