Add same padding mode for onnx (#456)

* fix pad calc

* add padding calc and test

* formatting

* made asym generic function

* formatting

Co-authored-by: mvermeulen <[email protected]>

src/onnx/onnx.cpp

@@ -16,6 +16,7 @@
 #include <migraphx/instruction.hpp>
 #include <migraphx/config.hpp>
 #include <migraphx/onnx.hpp>
+#include <migraphx/pad_calc.hpp>
 
 namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {
@@ -302,6 +303,24 @@ struct onnx_parser
         return curr_ins;
     }
 
+    template <class Op>
+    void check_asym_padding(instruction_ref& ins,
+                            std::vector<int64_t>& padding,
+                            Op& op,
+                            float pad_val = 0)
+    {
+        if(padding[0] != padding[2] || padding[1] != padding[3])
+        {
+            padding = {0, 0, padding[0], padding[1], 0, 0, padding[2], padding[3]};
+            ins     = prog.add_instruction(op::pad{padding, pad_val}, ins);
+        }
+        else
+        {
+            op.padding[0] = padding[0];
+            op.padding[1] = padding[1];
+        }
+    }
+
     instruction_ref parse_clip(const std::string&,
                                const attribute_map& attributes,
                                std::vector<instruction_ref> args)
@@ -424,7 +443,8 @@ struct onnx_parser
     parse_conv(const std::string&, attribute_map attributes, std::vector<instruction_ref> args)
     {
         Op op;
-        auto l0 = args[0];
+        auto l0      = args[0];
+        auto weights = args[1];
         if(contains(attributes, "pads"))
         {
             if(contains(attributes, "auto_pad"))
@@ -441,17 +461,7 @@ struct onnx_parser
             {
                 MIGRAPHX_THROW("padding should have 4 values");
             }
-            if(padding[0] != padding[2] || padding[1] != padding[3])
-            {
-                // insert zeros for pad op (args[0] has 4 dims)
-                padding = {0, 0, padding[0], padding[1], 0, 0, padding[2], padding[3]};
-                l0      = prog.add_instruction(op::pad{padding}, l0);
-            }
-            else
-            {
-                op.padding[0] = padding[0];
-                op.padding[1] = padding[1];
-            }
+            check_asym_padding(l0, padding, op);
         }
         if(contains(attributes, "strides"))
         {
@@ -471,7 +481,19 @@ struct onnx_parser
 
             if(s.find("SAME") != std::string::npos)
             {
-                op.padding_mode = op::padding_mode_t::same;
+                op.padding_mode                 = op::padding_mode_t::same;
+                std::vector<size_t> weight_dims = weights->get_shape().lens();
+                size_t weight_h                 = weight_dims[2];
+                size_t weight_w                 = weight_dims[3];
+
+                auto input_dims = l0->get_shape().lens();
+                std::vector<int64_t> padding(input_dims.size());
+                calculate_padding(
+                    0, padding, input_dims[2], op.stride[0], op.dilation[0], weight_h);
+                calculate_padding(
+                    1, padding, input_dims[3], op.stride[1], op.dilation[1], weight_w);
+
+                check_asym_padding(l0, padding, op);
             }
         }
         if(contains(attributes, "group"))
@@ -618,27 +640,10 @@ struct onnx_parser
             {
                 MIGRAPHX_THROW("PARSE_POOLING: padding should have 4 values");
             }
-            if(padding[0] != padding[2] || padding[1] != padding[3])
-            {
-                // insert zeros for pad op (args[0] has 4 dims)
-                padding = {0, 0, padding[0], padding[1], 0, 0, padding[2], padding[3]};
-                // MaxPool
-                if(op.mode == "max")
-                {
-                    l0 = prog.add_instruction(
-                        op::pad{padding, std::numeric_limits<float>::lowest()}, l0);
-                }
-                // AveragePool
-                else
-                {
-                    l0 = prog.add_instruction(op::pad{padding}, l0);
-                }
-            }
-            else
-            {
-                op.padding[0] = padding[0];
-                op.padding[1] = padding[1];
-            }
+            float pad_val = 0;
+            if(op.mode == "max")
+                pad_val = std::numeric_limits<float>::lowest();
+            check_asym_padding(l0, padding, op, pad_val);
         }
 
         if(contains(attributes, "strides"))

test/onnx/conv_autopad_same_test.onnx

@@ -0,0 +1,25 @@
+conv_autopad_same_test:�

+J
+
0
+
1
2"Conv*
+auto_pad"SAME�
*
+	dilations@
@
�
*
+strides@
@
�
conv_autopad_same_testZ
+
0
+

+

+
+ 
+ Z
+
1
+

+

+
+
+b
+
2
+

+

+

+ 
+ B

test/onnx/gen_onnx.py

@@ -492,6 +492,22 @@ def conv_autopad_fail_test():
     return ([node], [x, y], [out])
 
 
+@onnx_test
+def conv_autopad_same_test():
+    x = helper.make_tensor_value_info('0', TensorProto.FLOAT, [1, 3, 32, 32])
+    y = helper.make_tensor_value_info('1', TensorProto.FLOAT, [1, 3, 3, 3])
+    out = helper.make_tensor_value_info('2', TensorProto.FLOAT, [1, 1, 32, 32])
+
+    node = onnx.helper.make_node('Conv',
+                                 inputs=['0', '1'],
+                                 outputs=['2'],
+                                 dilations=[1, 1],
+                                 strides=[1, 1],
+                                 auto_pad='SAME')
+
+    return ([node], [x, y], [out])
+
+
 @onnx_test
 def conv_bias_test():
     x = helper.make_tensor_value_info('0', TensorProto.FLOAT, [1, 3, 32, 32])

test/onnx/onnx_test.cpp

@@ -341,6 +341,20 @@ TEST_CASE(conv_autopad_fail_test)
     EXPECT(test::throws([&] { optimize_onnx("conv_autopad_fail_test.onnx"); }));
 }
 
+TEST_CASE(conv_autopad_same_test)
+{
+    migraphx::program p;
+    auto l0 = p.add_parameter("0", {migraphx::shape::float_type, {1, 3, 32, 32}});
+    auto l1 = p.add_parameter("1", {migraphx::shape::float_type, {1, 3, 3, 3}});
+    migraphx::op::convolution op;
+    op.padding      = {1, 1};
+    op.padding_mode = migraphx::op::padding_mode_t::same;
+    p.add_instruction(op, l0, l1);
+
+    auto prog = optimize_onnx("conv_autopad_same_test.onnx");
+    EXPECT(p == prog);
+}
+
 TEST_CASE(conv_bias_test)
 {
     migraphx::program p;