[GitHub] [tvm] guberti commented on a diff in pull request #13242: [microTVM] Modernize Arm Cortex-M convolution schedules

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

guberti commented on code in PR #13242:
URL: https://github.com/apache/tvm/pull/13242#discussion_r1030627747


##########
python/tvm/relay/qnn/strategy/arm_cpu.py:
##########
@@ -0,0 +1,63 @@
+# 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.
+"""Quantized operator strategy for Arm CPU.
+
+As quantized op schedules, these are only used if the qnn.Legalize pass is disabled. The current
+schedules only work for fused operators with bias, as this is the most common use case. Only
+regular/depthwise conv2d is supported, but qnn_dense will be added eventually."""
+
+from tvm import topi, TVMError
+from .generic import qnn_conv2d_strategy
+from ... import op as _op
+from ...op.strategy.generic import is_depthwise_conv2d
+
+
+@qnn_conv2d_strategy.register("arm_cpu")
+def qnn_conv2d_strategy_arm_cpu(attrs, inputs, _out_type, target):
+    """qnn.conv2d strategy for Arm Cortex-M CPUs with DSP."""
+
+    if not (target.features.has_dsp and "cortex-m" in target.mcpu):
+        raise TVMError(
+            "Quantized Arm schedules only exist for Cortex-M with DSP! "
+            "The qnn.Legalize pass should be run for other Arm processors."
+        )
+
+    data = inputs[0]
+    kernel = inputs[1]
+    data_layout = attrs.data_layout
+    kernel_layout = attrs.kernel_layout
+    groups = attrs.groups
+    strategy = _op.OpStrategy()
+
+    if groups == 1:
+        if data_layout == "NHWC" and kernel_layout == "OHWI":

Review Comment:
   Good idea - I've written a summary and included it in the docstring for `qnn_conv2d_strategy_arm_cpu`.



-- 
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: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org