Fix wrapping handline
Several fixes to correctly handle memory wrapping:
- Bad in-place CP model for Conv and Pool (not enough margin, did not work with OpenMP);
- I/O offsets (usually named
iOffset
,oOffset
) must stay signed, otherwise memory wrapping won't work (@pineapple); -
INPUT_MEM_STRIDE
was used instead ofINPUT_MEM_CONT_SIZE
at some places; - Wrapping was not handled by outputs save and cmp (@axelfarr);
- Fixed several signed/unsigned comparison warnings;
- Use
size_t
instead ofint
when possible.
This MR fixes sporadic failure of CI due to wrong computation in lenet.py
with --mem_wrap
argument.
Edited by Olivier BICHLER
Merge request reports
Activity
Filter activity
changed milestone to %aidge v0.7.0
assigned to @olivierbichler
added Fix 🔥🔥 label
Please register or sign in to reply