SIMD Single Instruction, Multiple Data
CPU instruction that performs the same operation on multiple data elements in parallel — typically 4, 8, 16, or 32 at a time.
x86: MMX, SSE, SSE2-4, AVX, AVX2, AVX-512. ARM: NEON, SVE/SVE2. Drives multimedia codecs (H.264 encoders), 3D graphics (matrix multiplies), cryptography (AES-NI), JSON parsing (simdjson — 4 GB/s through AVX), and is the engine inside every numpy/Pandas operation. Different from GPU SIMT in that it runs on a CPU core. WebAssembly has a "WASM SIMD" extension for portable 128-bit operations.