Results from dpbench of PCA workload (running on a laptop with 11th Gen processor and Iris Xe graphics):
workload |
numpy |
dpnp CPU |
dpnp GPU |
size |
pca |
1.2 s |
3.6 s |
1.2 s |
1048576 |
demonstrates dpnp execution time on CPU which is 3 times greater than one of NumPy.