Skip to content

Conversation

kevinstephano
Copy link
Collaborator

  • Added fix to Reduction scheduler for Inner Dimension Reductions
  • Added a new reduction param: loop_unroll
  • Added an Executor method to compile fusions by string for debug
  • Fixed Scheduler tests to use randn() so large reductions were not matching on Infs.

…ction Scheduler to have TI level perf for FP16 inner dimension reductions. Fix tests to use randn() so large reductions aren't matching on inf.
Copy link
Collaborator

@jjsjann123 jjsjann123 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@@ -36,6 +36,31 @@ std::string FusionExecutor::getStructuredCode(const std::string& kernel) {
return code;
}

void FusionExecutor::compileFusionFromStr(
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are these only used for debugging? I don't see any real code using them. Pointing it out in case this is accidentally checked in.

@csarofeen csarofeen merged commit 0fbfa90 into 20_8_18_devel Aug 28, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants