The three files in this program both compare run times of various methods for matrix multiplication. tiledmm.c compares traditional matrix multiplication with tiled matrix multiplication, which decreases the size of the matrix multiplication into computation that can fit in the cache memory. threadmm.c compares tiled matrix multiplication with threaded matrix multiplication, where multipled threads simulatenously carry out the computation of the resultant matrix. mm.cu runs structured matrix multiplication on a GPU and compares it with an alternative CPU function. Running on the GPU, when tested in execution, is the most efficient method.
-
Notifications
You must be signed in to change notification settings - Fork 0
rubennoro/Matrix-Multiplication
About
This repo contains several matrix multiplication programs with different methods for improving efficiency such as tiling and synchronization.
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published