manupa-arm commented on a change in pull request #8019:
URL: https://github.com/apache/tvm/pull/8019#discussion_r631943300
##########
File path: src/tir/analysis/calculate_workspace.cc
##########
@@ -46,6 +48,10 @@ size_t WorkspaceCalculator::operator()(const PrimFunc& func)
{
return this->max_size;
}
+size_t WorkspaceCalculator::GetByteAlignedSize(size_t non_aligned_size) {
+ return ((non_aligned_size + byte_alignment - 1) / byte_alignment) *
byte_alignment;
Review comment:
Yes, that is right -- in that work we'll just directly encode the
offsets within the buffer that can respect the alignment. Until that we can
have a global (sub-optimal) alignment accross all tensors accessed. Does that
make sense?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
