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
As per the discussion with @cbauinge, it would be nice to pre-allocate all the
necessary memory for IFGF operator so that the foward map is allocation-free.
To do this, we have to:
Find the largest number interpolation coefficients needed at any level of the
source tree
Pre-allocate two buffers of that size to keep the interpolation
values/coefficients of two consecutive levels.
If properly done, this should allow (I think) for an allocation-free forward
map. The other advantage is that we can have an a priori upper bound on the maximum
memory required, and so will not run out of memory during the computation of a
matrix/vector product.
The text was updated successfully, but these errors were encountered:
As per the discussion with @cbauinge, it would be nice to pre-allocate all the
necessary memory for
IFGF
operator so that the foward map is allocation-free.To do this, we have to:
source tree
values/coefficients of two consecutive levels.
If properly done, this should allow (I think) for an allocation-free forward
map. The other advantage is that we can have an a priori upper bound on the maximum
memory required, and so will not run out of memory during the computation of a
matrix/vector product.
The text was updated successfully, but these errors were encountered: