Interview
summary
Explains using CPU intrinsics from C# where the JIT replaces calls with direct machine instructions, essentially inline assembly in C# that targets array-style workloads; claims most intrinsics are about vector work — 96 percent deal with vectorization — and demonstrates it by rewriting Array.Sort’s quicksort piece as a hybrid sorter for tiny arrays, building a drop-in replacement as a NuGet candidate that preserves correctness while cutting per-element work with explicit SIMD instructions.
AI-written summary · about
related talks
52mAdvanced SIMD Algorithms in Pictures
01h 05mThe Most Important Optimizations to Apply in Your C++ Programs
42mFaster programs with your compilers autovectorization feature
01h 00mJust-in-Time Compilation
45mFreaking computers, how do they work?
57mA Long Journey of Changing std::sort Implementation at Scale
01h 00mIntroduction to Hardware Efficiency in Cpp
01h 08mAn implementer's view of Boost.SIMD