elvin-n commented on code in PR #11357:
URL: https://github.com/apache/tvm/pull/11357#discussion_r890927872
##########
include/tvm/te/operation.h:
##########
@@ -76,6 +78,11 @@ class TVM_DLL OperationNode : public Object {
* \return type of i-th output.
*/
virtual DataType output_dtype(size_t i) const = 0;
+ /**
+ * @brief Returns memory scope of operation
+ * TODO(amalyshe): add support for individual output's tensors, not the only
one
+ */
+ String memory_scope() const { return memory_scope_; }
Review Comment:
Yes, this can be used and I will try to implement passing of memory_scope
through bindings.
But this is not consistent because by fact we have virtual device / memory
scope for relay level, we have virtual device / memory scope for tir level, but
right now we do not have any landing of virtual devices from relay to tir.
While we have a converting of relay entities to tir through lowering mechanism.
This is wrong from architecture point of view. We need to take into account
all relay attributes during conversion to tir.
--
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]
