[GitHub] [incubator-tvm] kevinthesun commented on a change in pull request #5962: [Ansor][AutoTVM v2.0] Part 0: Ansor minimum system for auto schedule generating

[GitBox <gi...@apache.org>]

kevinthesun commented on a change in pull request #5962:
URL: https://github.com/apache/incubator-tvm/pull/5962#discussion_r449235073



##########
File path: python/tvm/ansor/auto_schedule.py
##########
@@ -0,0 +1,206 @@
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements.  See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership.  The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License.  You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied.  See the License for the
+# specific language governing permissions and limitations
+# under the License.
+
+"""
+User interface for Ansor auto-scheduler.
+
+The basic schedule search process for Ansor is designed to be:
+`Program sampling` -> `Performance Tuning`.
+
+In `Program sampling`, we use some predefined or heuristic rules to generate several initial
+schedules. Based on these initial start points, we have `Performance Tuning` to apply cost model
+and evolutionary search to seek for schedules with the best performance. Candidate schedules will
+be measured in the target hardware.
+"""
+
+import tvm._ffi
+from tvm.runtime import Object
+from .compute_dag import ComputeDAG
+from .measure import LocalBuilder, LocalRunner
+from . import _ffi_api
+
+
+@tvm._ffi.register_object("ansor.HardwareParams")
+class HardwareParams(Object):
+    """ The parameters of target hardware, this is used to guide the search process of
+    SearchPolicy.
+
+    TODO(...): This is considering to merge with the new Target:
+    https://discuss.tvm.ai/t/rfc-tvm-target-specification/6844
+
+    Parameters

Review comment:
       How do we decide these args? Some of them can be auto-detected from OS but what about ```max_unroll_vec``` and ```max_innermost_split_factor```?




----------------------------------------------------------------
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:
users@infra.apache.org