This is an automated email from the ASF dual-hosted git repository.
masahi pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
The following commit(s) were added to refs/heads/main by this push:
new 636463d16c Consider pad value and input zero point in
FoldExplicitPading (#11127)
636463d16c is described below
commit 636463d16c8f1713a3d93793b60d21dde9b6a6f7
Author: ibsidorenko <[email protected]>
AuthorDate: Fri May 13 22:58:14 2022 +0300
Consider pad value and input zero point in FoldExplicitPading (#11127)
This commit adds the following:
Do not fold `nn.pad` and `qnn.conv2d` if padding value is not
equal to input zero point of qnn operation. Added unit test
to check such behaviour.
---
src/relay/transforms/fold_explicit_padding.cc | 15 +++++++----
.../relay/test_pass_fold_explicit_padding.py | 29 ++++++++++++++++++++++
2 files changed, 39 insertions(+), 5 deletions(-)
diff --git a/src/relay/transforms/fold_explicit_padding.cc
b/src/relay/transforms/fold_explicit_padding.cc
index 6aac995e35..60b52c170a 100644
--- a/src/relay/transforms/fold_explicit_padding.cc
+++ b/src/relay/transforms/fold_explicit_padding.cc
@@ -129,22 +129,27 @@ class SimplifyConvPad {
ICHECK(pad_node);
const PadAttrs* param = pad_node->attrs.as<PadAttrs>();
ICHECK(param);
- Array<Expr> args = pad_node->args;
auto x = node_map[x_][0];
auto w = node_map[w_][0];
// Possibly perform more optimizations if the pad_value is 0
- const ConstantNode* pad_value = args[1].as<ConstantNode>();
+ const Expr& pv = pad_node->args[1];
+ const ConstantNode* pad_value = pv.as<ConstantNode>();
if (node_map.find(qconv2d_) != node_map.end()) {
Attrs attrs = GetAttrs(param, call_node->attrs.as<Conv2DAttrs>());
auto input_zero_point = node_map[input_zero_point_][0];
auto kernel_zero_point = node_map[kernel_zero_point_][0];
auto input_scale = node_map[input_scale_][0];
auto kernel_scale = node_map[kernel_scale_][0];
- return Call(call_node->op,
- {x, w, input_zero_point, kernel_zero_point, input_scale,
kernel_scale}, attrs,
- call_node->type_args, call_node->span);
+ // Fold Padding and QNN Convolution only if pad value == input zero
point.
+ if (IsEqualScalar(input_zero_point, pv)) {
+ return Call(call_node->op,
+ {x, w, input_zero_point, kernel_zero_point, input_scale,
kernel_scale}, attrs,
+ call_node->type_args, call_node->span);
+ } else {
+ return post;
+ }
} else if (param->pad_mode == "constant" && pad_value &&
ToScalar(pad_value->data) == 0.0) {
Attrs attrs;
if (node_map.count(conv1d_)) {
diff --git a/tests/python/relay/test_pass_fold_explicit_padding.py
b/tests/python/relay/test_pass_fold_explicit_padding.py
index 48b5e510d0..2887c0774b 100644
--- a/tests/python/relay/test_pass_fold_explicit_padding.py
+++ b/tests/python/relay/test_pass_fold_explicit_padding.py
@@ -144,6 +144,35 @@ def fold_pad_qconv2d():
assert tvm.ir.structural_equal(a, b, map_free_vars=True), "Actual = \n" +
str(a)
+def test_pad_qconv2d_no_fold():
+ def get_expr():
+ x = relay.var("x", shape=(1, 1, 2, 2), dtype="int8")
+ weight = relay.var("weight", shape=(1, 1, 2, 2), dtype="int8")
+ # Pad value and input zp are not equal
+ pad_value = 1
+ input_zero_point = 0
+ pad = relay.nn.pad(x, [[0, 0], [0, 0], [1, 1], [1, 1]],
pad_value=pad_value)
+ return relay.qnn.op.conv2d(
+ pad,
+ weight,
+ relay.const(input_zero_point, "int32"),
+ relay.const(0, "int32"),
+ relay.const(1, "float32"),
+ relay.const(1, "float32"),
+ channels=1,
+ kernel_size=(2, 2),
+ padding=(0, 0),
+ )
+
+ a = run_opt_pass(get_expr(), relay.transform.FoldExplicitPadding())
+ b = run_opt_pass(get_expr(), transform.InferType())
+
+ assert tvm.ir.structural_equal(a, b, map_free_vars=True), (
+ "\nActual = \n" + str(a) + "\nExpected = \n" + str(b)
+ )
+
+
if __name__ == "__main__":
test_simplify_conv_pad()
fold_pad_qconv2d()
+ test_pad_qconv2d_no_fold()
