This project was a research about speeding up vector math operations using SSE instruction set.

Project focused on boosting efficiency of math operations on vectors and matrices using the x86 SIMD unit. Selected math operations were implemented in C language using CPU/FPU for computations and then tested against the version implemented in assembly using SIMD.

The chart below shows computation timings in microseconds for FPU and SSE versions of algorithms, where 2000 vector or matrix elements were processed. Intel Core2 Duo T9300 @ 2,5 GHz was used for this benchmark.

Project was divided into two parts: ASM math code exported into DLL and benchmark module with C math implementations. Source code can be downloaded >here<