tqchen commented on code in PR #18604:
URL: https://github.com/apache/tvm/pull/18604#discussion_r2643058305
##########
src/runtime/cuda/cuda_module.cc:
##########
@@ -173,10 +173,12 @@ class CUDAWrappedFunc {
public:
// initialize the CUDA function.
void Init(CUDAModuleNode* m, ObjectPtr<Object> sptr, const std::string&
func_name,
- size_t num_void_args, const std::vector<std::string>&
launch_param_tags) {
+ size_t num_void_args, const std::vector<std::string>&
launch_param_tags,
+ bool has_programmatic_dependent_launch) {
Review Comment:
thanks a lot for the contribution, to align with existing impl, we can use
launch param tag, see LaunchParamConfig implementation. The launch param tag
can be specified as part of `tir.kernel_launch_params`, which then set the
LaunchPramConfig setting
we can use
constexpr const char* kUseProgramaticDependentLaunch =
"tir.use_programtic_dependent_launch";
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
