You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Feb 3, 2024. It is now read-only.
-14:43: initial version => 11,16 sec (10cb14c46c0474eeee598f1802ac5b6d875f457c)
-14:45: The materialization is done less times, but I have to use shared pointers. => 11.026 sec (b0806e82cb06475bcf6b48cbaf1281dda428f2de)
16 February
-16:54: removed all virtual calls from OptimizableMD.get() => 1.88 sec
18 february
-10:26: not calling optimize on children of Concatenate matrix => 1.5 sec (16b959d3e56b27ad5f59f02160fb540968a99293)
-11:55: using std::async => 1.2 sec (91dcef68a65a3c702b8a74f44456edc6e58d2c9d)
-12:47: using -O3. Not comparable to the results above, but further benchmarks will be more indicative or real world performances => 0.09 sec (0812e089a8a796f76dfd6b98d6bf66ec5e486ddf)
20 february
-18:01: Base new test => 0.475 sec
-18:05: Calling optimize before get() for every MatrixData => 0.400 sec
-20:30: New test => 3.06 sec
22 February
-14:57: Using laptop, test took => 5.14 sec
-15:06: Vectors are freed when not needed anymore. Slower, but needs less memory => 5.48 sec
-15:44 BaseMultiplication caches pointers to left and right => 5.30 sec