Closed oltolm closed 4 years ago
I used cv2pdb with dxvk and it was extremely slow. I introduced a symbol cache which drastically reduces the run time.
cv2pdb
dxvk
Benchmark results:
LGTM. This also solves the issue that if a symbol and the same symbol with an underscore both exist, the wrong one might be chosen so far. Now the perfect match is used if it exists. Thanks for your contribution.
I used
cv2pdb
withdxvk
and it was extremely slow. I introduced a symbol cache which drastically reduces the run time.Benchmark results: