Lunderberg commented on a change in pull request #9727:
URL: https://github.com/apache/tvm/pull/9727#discussion_r818725019
##########
File path: include/tvm/tir/builtin.h
##########
@@ -105,10 +105,10 @@ TVM_DLL const Op& large_uint_imm();
TVM_DLL const Op& q_multiply_shift();
/*!
- * \brief See pesudo code
+ * \brief See pseudo code
*
- * Handle address_of(Load *op) {
- * return &op->buffer_var[index];
Review comment:
I think it needs to allow non-flat indices, both for TIR prior to
flattening (N-d indices for N-d logical shape of buffer) and for TIR
post-flattening on supported targets (N-d indices for N-d physical shape).
I've added to the docstring that the buffer address must be compatible with the
targets used.
--
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]
