-
Notifications
You must be signed in to change notification settings - Fork 3.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[BYOC] InlineCompilerFunctions helper pass #11923
Conversation
The TensorRT BYOC integration needs to 'undo' partitionings in some situations. Add an InlineCompilerFunctions pass to make that robust. In particular, it must undo both the 'partitioning' (ie separating out the "Compiler" function) and any 'compositing' (ie separating out small sub-graphs as "Composite" functions). Fix misspelled nn.bias_add while there. Note that the current implementation is broken but untested in CI. I have all the tests fixed in a follow-up PR.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
cc @comaniac
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The changes look clean. Thanks!
One miner point is the name InlineCompilerFunctions
seems a bit confusing although the implementation itself is general to inline functions with their global vars specified. Would that be better to just call something like InlineFunctions
?
Thanks @comaniac, agree -- the iface changed but I forgot to update the pass name. |
* [BYOC] InlineCompilerFunctions helper pass The TensorRT BYOC integration needs to 'undo' partitionings in some situations. Add an InlineCompilerFunctions pass to make that robust. In particular, it must undo both the 'partitioning' (ie separating out the "Compiler" function) and any 'compositing' (ie separating out small sub-graphs as "Composite" functions). Fix misspelled nn.bias_add while there. Note that the current implementation is broken but untested in CI. I have all the tests fixed in a follow-up PR. * - Lints * - Only AOT compilation paths ensure "executor" is provided as a Target attribute.
* [BYOC] InlineCompilerFunctions helper pass The TensorRT BYOC integration needs to 'undo' partitionings in some situations. Add an InlineCompilerFunctions pass to make that robust. In particular, it must undo both the 'partitioning' (ie separating out the "Compiler" function) and any 'compositing' (ie separating out small sub-graphs as "Composite" functions). Fix misspelled nn.bias_add while there. Note that the current implementation is broken but untested in CI. I have all the tests fixed in a follow-up PR. * - Lints * - Only AOT compilation paths ensure "executor" is provided as a Target attribute.
The TensorRT BYOC integration needs to 'undo' partitionings in some situations. Add an
InlineCompilerFunctions pass to make that robust. In particular, it must undo both the
'partitioning' (ie separating out the "Compiler" function) and any 'compositing' (ie separating
out small sub-graphs as "Composite" functions).
Fix misspelled nn.bias_add while there.
Note that the current implementation has been broken since #11203 (mea culpa) but was untested by CI. I have all the tests fixed in a follow-up PR and a few other bug fixes.