added basic computations; builds but doesn't pass test

src/include/migraphx/op/resize.hpp

@@ -18,12 +18,77 @@ namespace migraphx {
 inline namespace MIGRAPHX_INLINE_NS {
 namespace op {
 
+   // from parse_resize.cpp
+auto& get_nearest_op(const std::string& near_mode)
+{
+    using nearest_op = std::function<std::size_t(std::size_t, double)>;
+    static std::unordered_map<std::string, nearest_op> const nearest_ops = {
+        {"round_prefer_floor",
+        [=](std::size_t d_in, double val) {
+            val = std::max(0.0, std::min(d_in - 1.0, val));
+            return static_cast<std::size_t>(std::ceil((val - 0.5)));
+        }},
+        {"round_prefer_ceil",
+        [=](std::size_t d_in, double val) {
+            val = std::max(0.0, std::min(d_in - 1.0, val));
+            return static_cast<std::size_t>(std::round((val)));
+        }},
+        {"floor",
+        [=](std::size_t d_in, double val) {
+            val = std::max(0.0, std::min(d_in - 1.0, val));
+            return static_cast<std::size_t>(std::floor((val)));
+        }},
+        {"ceil", [=](std::size_t d_in, double val) {
+            val = std::max(0.0, std::min(d_in - 1.0, val));
+            return static_cast<std::size_t>(std::ceil((val)));
+        }}};
+
+    if(not contains(nearest_ops, near_mode))
+    {
+        MIGRAPHX_THROW("RESIZE: nearest_mode " + near_mode + " not supported!");
+    }
+
+    return nearest_ops.at(near_mode);
+}
+
+const auto& get_original_idx_op(const std::string& mode)
+{
+    using original_idx_op = std::function<double(std::size_t, std::size_t, std::size_t, double)>;
+    static std::unordered_map<std::string, original_idx_op> const idx_ops = {
+        {"half_pixel",
+         [=](std::size_t, std::size_t, std::size_t idx, double scale) {
+             return (idx + 0.5) / scale - 0.5;
+         }},
+        {"pytorch_half_pixel",
+         [=](std::size_t, std::size_t l_out, std::size_t idx, double scale) {
+             return l_out > 1 ? (idx + 0.5) / scale - 0.5 : 0.0;
+         }},
+        {"align_corners",
+         [=](std::size_t l_in, std::size_t l_out, std::size_t idx, double) {
+             return (l_out == 1) ? 0.0 : (1.0 * idx * (l_in - 1.0) / (l_out - 1.0));
+         }},
+        {"asymmetric",
+         [=](std::size_t, std::size_t, std::size_t idx, double scale) { return idx / scale; }},
+        {"tf_half_pixel_for_nn", [=](std::size_t, std::size_t, std::size_t idx, double scale) {
+             return (idx + 0.5) / scale;
+         }}};
+
+    if(not contains(idx_ops, mode))
+    {
+        MIGRAPHX_THROW("RESIZE: coordinate_transformation_mode " + mode + " not supported!");
+    }
+
+    return idx_ops.at(mode);
+}
+
 struct resize
 {
     // TODO:   indicators.  The real scales and sizes are inputs, not attributes.
     std::vector<float> scales;
     std::vector<int64_t> sizes;
-    int mode = 0; // 1: nereast 2: bilinear/linear 3: cubic
+    std::string nearest_mode;
+
+    int mode = 0; // 1: nearest 2: bilinear/linear 3: cubic
     std::string coordinate_transformation_mode;
 
     std::string name() const { return "resize"; }
@@ -33,10 +98,13 @@ struct resize
     {
         return pack(f(self.scales, "scales"),
                     f(self.sizes, "sizes"),
+                    f(self.nearest_mode,"nearest_mode"),
                     f(self.mode,"mode"),
                     f(self.coordinate_transformation_mode,"coordinate_transformation_mode"));
     }
 
+ 
+
     shape compute_shape(std::vector<shape> inputs) const
     {
         // check_shapes{{inputs[0]}, *this, true}.has(2);
@@ -77,32 +145,6 @@ struct resize
         return {inputs.front().type(), dyn_dims};
 
        
-        // static input.  
-        // if(!scales.empty())
-        // {
-        //     // 计算输出blob大小
-        //     auto in_s    = inputs[0];
-        //     auto in_lens = in_s.lens();
-        //     if(in_lens.size() != scales.size())
-        //     {
-        //         MIGRAPHX_THROW("PARSE_UPSAMPLE: ranks of input and scale are different!");
-        //     }
-        //     std::vector<std::size_t> out_lens(in_lens.size());
-        //     std::transform(in_lens.begin(),
-        //                 in_lens.end(),
-        //                 scales.begin(),
-        //                 out_lens.begin(),
-        //                 [&](auto idx, auto scale) { return static_cast<std::size_t>(idx * scale); });
-            
-        //     return shape{in_s.type(), out_lens};
-        // }
-        // else if(!sizes.empty())
-        // {
-        //     return shape{inputs[0].type(), sizes};
-
-        // }
-
-        
     }
 
     argument compute(const dyn_output& dyn_out, std::vector<argument> args) const
@@ -110,17 +152,21 @@ struct resize
         // See scatter.hpp or gather.hpp for how to do a similar iteration with reduction
                 // iterate through items in shape
         argument result{dyn_out.computed_shape};
-        // negative axis means counting dimensions from back
-        auto lens                 = args[0].get_shape().lens();
-//Everything that follows is placeholder logic
-        auto axis = 2;
-        std::size_t axis_dim_size = lens[axis];
+        auto nearest_op = get_nearest_op(nearest_mode);
+        auto idx_op              = get_original_idx_op(coordinate_transformation_mode);
+
+        auto in_lens                 = args[0].get_shape().lens();
+        auto out_lens = dyn_out.computed_shape.lens();
+
+        // temp.  This is a placeholder for reading the desired dimensions or scale
+        std::vector<double> vec_scale={1., 1., 5./3., 8./3.};
+
        // max dimension in axis
         visit_all(result, args[0])([&](auto output, auto data) {
 
             // the size input
             args[1].visit([&](auto indices) {
-for(auto aa : indices ) std::cout << aa << "   indices \n";                
+                for(auto aa : indices ) std::cout << aa << "   indices \n";                
                 if(dyn_out.computed_shape.scalar())
                 {
                     std::cout << " scalar output\n";
@@ -130,21 +176,29 @@ for(auto aa : indices ) std::cout << aa << "   indices \n";
                     // for each element in output, calculate index in input
                     for(auto bb : data) std::cout << bb << "   zzz data \n";
 
-                    // auto out_lens  = data.get_shape().lens();
-                    // out_lens[axis] = indices.get_shape().elements();
                     migraphx::shape out_comp_shape{data.get_shape().type(), indices};
 
                     shape_for_each(out_comp_shape, [&](const auto& out_idx_v, size_t out_idx) {
-                        auto data_idx   = out_idx_v;
-                        auto in_index   = indices[data_idx[axis]];
-                        in_index        = (in_index < 0) ? in_index + axis_dim_size : in_index;
-                        data_idx[axis]  = in_index;
-                        output[out_idx] = data(data_idx.begin(), data_idx.end());
-                        std::cout << " !!!!! did something\n";
+                        // Show the output indices.  Last index iterates fastest
+                        for(auto vv : out_idx_v ) std::cout << vv << "    ";         
+                        std::cout <<out_idx <<  "  out_index\n"; 
+                        std::cout << nearest_mode << "\n";               
+
+                        // populate output at this index
+                            // output[out_idx] = data(data_idx.begin(), data_idx.end());
+                        std::vector<size_t> in_idx(out_idx_v.size());
+                        for(auto ii = 0; ii < out_idx_v.size(); ++ii)
+                        {
+                            auto idx_val = idx_op(in_lens[ii], out_lens[ii], out_idx_v[ii], vec_scale[ii]);
+                            std::cout << in_lens[ii] << " " <<  out_lens[ii] << " " <<  out_idx_v[ii] << " " <<  vec_scale[ii] << "==> " <<  idx_val << "\n";
+                            in_idx[ii]   = nearest_op(in_lens[ii], idx_val);
+                        }                  
+                        // output[out_idx] = data.at(in_idx);
                     });
                 }
             });
         });
+        std::cout << " finish resize\n";
         return result;
     }
 

test/ref/resize.cpp

@@ -46,9 +46,11 @@ TEST_CASE(resize_test_1)
     auto a1  = mm->add_literal(migraphx::literal{size_input, size_values});
 
 
-    mm->add_instruction(migraphx::make_op("resize", {{"sizes", {1}}, {"scales", {}}}), a0, a1);
+    mm->add_instruction(migraphx::make_op("resize", {{"sizes", {1}}, {"scales", {}}, {"nearest_mode", "floor"}
+      , {"coordinate_transformation_mode", "half_pixel"}}), a0, a1);
     p.compile(migraphx::make_target("ref"));
     auto result = p.eval({}).back();
+
     std::vector<float> res_data(4 * 5);
     std::vector<float> golden = {0.5f, 1.5f, 2.5f, 6.5f, 7.5f, 8.5f};
     result.visit([&](auto output) { res_data.assign(output.begin(), output.end()); });