Glossary
A
AVX (Advanced Vector Extensions): 256-bit SIMD instruction set for x86_64 CPUs (Sandy Bridge+, 2011+).
AVX2: Enhanced version of AVX with FMA (Haswell+, 2013+).
AVX-512: 512-bit SIMD instruction set (Zen 4, Sapphire Rapids+, 2022+).
B
Backend: Implementation executing vector operations (Scalar, SSE2, AVX2, GPU).
Backend Equivalence: All backends produce identical results.
C
CPU Feature Detection: Runtime SIMD detection using is_x86_feature_detected!().
Criterion.rs: Statistical benchmarking framework for Rust.
E
Element-wise Operation: Operation on each element independently (add, mul).
EXTREME TDD: Test methodology with >90% coverage, mutation testing.
F
FMA (Fused Multiply-Add): Instruction computing a * b + c.
G
GPU (Graphics Processing Unit): Massively parallel compute processor.
N
NEON: 128-bit SIMD for ARM64 CPUs.
S
SIMD (Single Instruction Multiple Data): Parallel execution on multiple elements.
SSE2: 128-bit SIMD baseline for x86_64.
W
WASM (WebAssembly): Portable bytecode for browsers.
wgpu: Rust library for GPU compute.