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
Yes. Nevertheless, the parsing of 1GB floating point values from plain text files is more expensive than the actual computation of subsequence Euclidean Distance. Same goes for efficiently lower-bounded DTW.
Yes and No. The length limitation is caused by the limitation of constant memory on the GPU. Writing the query and warping envelope to (slower) global memory will fix that. Nevertheless, assuming 10% relative warping window for the Sakoe-Chiba constraint, query sizes above 10K will most likely not be possible due to limitations of shared memory. Moving the penalty matrix to (slow) global memory would fix that at the expense of speed.
These are not issues just some suggestions:
The text was updated successfully, but these errors were encountered: