tkonolige commented on a change in pull request #8975:
URL: https://github.com/apache/tvm/pull/8975#discussion_r705512883
##########
File path: python/tvm/contrib/debugger/debug_executor.py
##########
@@ -68,7 +73,7 @@ def create(graph_json_str, libmod, device, dump_root=None):
"Please set '(USE_PROFILER ON)' in " "config.cmake and rebuild TVM
to enable debug mode"
)
func_obj = fcreate(graph_json_str, libmod, *device_type_id)
- return GraphModuleDebug(func_obj, dev, graph_json_str, dump_root)
+ return GraphModuleDebug(func_obj, dev, graph_json_str, dump_root,
iters_num, repeat)
Review comment:
Can you also add `min_repeat_ms` to the parameters. It is the 3rd
parameter passed to `self.run_individual`.
##########
File path: python/tvm/contrib/debugger/debug_executor.py
##########
@@ -93,9 +98,16 @@ class GraphModuleDebug(graph_executor.GraphModule):
dump_root : str
To select which folder the outputs should be kept.
None will make a temp folder in /tmp/tvmdbg<rand_string> and does the
dumping
+ iters_num : int
Review comment:
With `time_evaluator` and `GraphExecutor.benchmark`, this parameter is
called `number`. It is not that descriptive, but I think it would be worthwhile
to keep the parameter names the same.
Also, you can maybe just copy and paste the description of `number` and
`repeat` from `graph_executor.py`?
--
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]
