From the archive
Wednesday, 28 October
Folded from 11 sources that day.
Sorted by quality and by how many sources covered it, the same for everyone.
-
Making SIMD Code Scale to Any Vector Width
Shows how to transpose vec3/vec4 math into SIMD lanes so N dot products and frustum culling scale to 8 or 16 wide, with benchmarks on a Ryzen 7 5800 under Clang.
-
C++26: std::indirect
C++26 adds std::indirect (P3019R14), a vocabulary type giving value semantics to heap-allocated objects with deep copying and const propagation, without std::unique_ptr boilerplate.
-
Billion-query-per-second lookup tables in C++
Shows how to build compile-time perfect hash maps for fixed string keys in C++23, following gperf's design, with no external tool and bounded constexpr work.