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This repository has been archived by the owner on Nov 1, 2024. It is now read-only.
Summary:
Pull Request resolved: pytorch/torchrec#177
Perf units are calculated with bandwidth being in terms if GB/sec. To accurately model perfs we want to put it in terms of bytes/ms
Reviewed By: dstaay-fb
Differential Revision: D34758134
fbshipit-source-id: 46a0560ec37972f92a60dd78024bb6b949e702dc
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
We should remove the
_offset
and_length
inBaseColumn
:https://github.com/facebookresearch/torcharrow/blob/d680bfdc0f6a6bb6c3a29c2a67d62006782d6558/csrc/velox/column.h#L223-L224
There are multiple places where we do not properly track this, such as in expression evaluation:
https://github.com/facebookresearch/torcharrow/blob/d680bfdc0f6a6bb6c3a29c2a67d62006782d6558/csrc/velox/column.cpp#L236-L238
We should be able to not track these in the
BaseColumn
anymore without losing any functionality.We also may want to support UDF evaluation for different offsets, such as:
Slicing the vector with the BufferView might be the right solution.
cc: @wenleix
The text was updated successfully, but these errors were encountered: