[PATCH] D80051: [OpenMPOpt] Test case 1 - Latency Hiding for Host to Device Memory Transfers

[Johannes Doerfert via Phabricator via llvm-commits]

jdoerfert added inline comments.


================
Comment at: llvm/test/Transforms/OpenMP/mem_transfer_hiding.ll:2
+; RUN: opt -S -openmpopt -stats < %s 2>&1
+; REQUIRES: asserts
+
----------------
jdoerfert wrote:
> hamax97 wrote:
> > jdoerfert wrote:
> > > Why do you want to run this with stats? I guess later you just use filecheck and verify the IR, right?
> > I thought it would be useful to perhaps count the number of split runtime calls, and how much they were moved forward and backwards. What do you think?. I have no problem not doing it.
> We will eventually check the resulting IR which contains all the information. We basically compare against the expected splits and movements so any derivation needs to be flagged. That said, having a test case to verify our statistic tracking works and one to verify the remarks (=feedback to the user) work, is important. Though, both should be separate test cases with appropriate names.
No error output and asserts needed. We should however already use filecheck to check for the outut we care about. You can use `llvm/utils/update_test_checks.py` on this file. Once we optimize this we can nicely see the effect in the diff.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D80051/new/

https://reviews.llvm.org/D80051