adstraw commented on code in PR #12785:
URL: https://github.com/apache/tvm/pull/12785#discussion_r971383716
##########
src/driver/driver_api.cc:
##########
@@ -225,6 +225,7 @@ Array<tvm::transform::Pass> CreatePassList(bool
disable_loop_partition) {
}
// LowerVtcmAlloc must occur after any transformations that modify memory
allocation locations
pass_list.push_back(tir::transform::LowerVtcmAlloc());
+ pass_list.push_back(tir::transform::LowerAsyncDMA());
Review Comment:
Also realized that I may need to have a device specific (Hexagon) enable
here and/or TIR annotation or some other mechanism to enable this pass to avoid
breaking existing async pipeline implementations e.g. cuda.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
