zhiics commented on a change in pull request #4723: [Relay] Invoke tvm::build
from relay compile_engine and interpreter
URL: https://github.com/apache/incubator-tvm/pull/4723#discussion_r367756167
##########
File path: src/relay/backend/compile_engine.cc
##########
@@ -599,12 +599,13 @@ class CompileEngineImpl : public CompileEngineNode {
CCacheValue value = LowerInternal(key);
if (value->packed_func != nullptr) return value->packed_func;
// build the function.
+ tvm::runtime::Module m;
if (const auto* f = runtime::Registry::Get("relay.backend.build")) {
- tvm::runtime::Module m = (*f)(value->cached_func->funcs, key->target);
- value->packed_func = m.GetFunction(value->cached_func->func_name);
+ m = (*f)(value->cached_func->funcs, key->target);
} else {
- LOG(FATAL) << "relay.backend.build is not registered";
+ m = build(value->cached_func->funcs, key->target, key->target,
BuildConfig::Current());
Review comment:
according to the syntax we had in Python, should we do the following?
```c++
m = build(value->cached_func->funcs, key->target, Target(nullptr),
BuildConfig::Current());
```
Same to the one in intepreter.cc
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
