AVX2

A 256-bit SIMD instruction set expansion that doubles integer throughput and introduces FMA3 for accelerated vector processing.

AVX2 (Advanced Vector Extensions 2) launched with Intel's Haswell architecture in 2013 to expand 128-bit integer SIMD instructions to full 256-bit widths. It adds critical features: Fused Multiply-Add (FMA3) for high-precision math and gather support for fetching non-adjacent data from memory. This instruction set is the standard for performance-critical software: 4K video encoding (x265), scientific simulations, and cryptographic hashing. By handling 32 bytes of data per instruction, AVX2 provides a significant efficiency ceiling over legacy SSE4.2 implementations.

https://www.intel.com/content/www/us/en/developer/articles/technical/intel-advanced-vector-extensions-2-intel-avx2-instructions.html