Unify Convolution FWD XDL 1D/2D implementation. (#93)

* Convolution ND * Code unification across dimensions for generating tensor descriptors. * Example * Instances * Move convnd f32 instance file to comply with repo structure. * Conv 1D tensor layouts. * Formatting and use ReferenceConv * Reference ConvFwd supporting 1D and 2D convolution. * Debug printing TensorLayout name. * Conv fwd 1D instance f32 * Refactor conv ND example. Needed to support various conv dimensio. Needed to support various conv dimensions * Rename conv nd example director to prevent conflicts. * Refactor some common utility to single file. Plus some tests. * Refactor GetHostTensorDescriptor + UT. * Add 1D test case. * Test reference convolution 1d/2d * Remove some leftovers. * Fix convolution example error for 1D * Refactor test check errors utility function. * Test Conv2D Fwd XDL * More UT for 1D case. * Parameterize input & weight initializers. * Rename example to prevent conflicts. * Split convnd instance into separate files for 1d/2d * Address review comments. * Fix data type for flops/gbytes calculations. * Assign example number 11. Co-authored-by: Adam Osewski <aosewski@amd.com> Co-authored-by: Chao Liu <chao.liu2@amd.com> [ROCm/composable_kernel commit: 756a761727]
2026-05-16 02:54:21 +00:00 · 2022-02-23 17:44:20 +01:00
parent c2e3fa5c91
commit a2386b6bbf
17 changed files with 2698 additions and 108 deletions
--- a/test/include/test_util.hpp
+++ b/test/include/test_util.hpp
@@ -0,0 +1,84 @@
+#ifndef TEST_UTIL_HPP
+#define TEST_UTIL_HPP
+
+#include <cmath>
+#include <cstdlib>
+#include <iostream>
+#include <iomanip>
+#include <limits>
+#include <type_traits>
+#include <vector>
+
+namespace test_util {
+
+template <typename T>
+typename std::enable_if<std::is_floating_point<T>::value, bool>::type
+check_err(const std::vector<T>& out,
+          const std::vector<T>& ref,
+          const std::string& msg,
+          T rtol = static_cast<T>(1e-5),
+          T atol = static_cast<T>(1e-8))
+{
+    if(out.size() != ref.size())
+    {
+        std::cout << "out.size() != ref.size(), :" << out.size() << " != " << ref.size()
+                  << std::endl
+                  << msg << std::endl;
+        return false;
+    }
+
+    bool res{true};
+    int err_count = 0;
+    T err         = 0;
+    T max_err     = std::numeric_limits<T>::min();
+    for(std::size_t i = 0; i < ref.size(); ++i)
+    {
+        err = std::abs(out[i] - ref[i]);
+        if(err > atol + rtol * std::abs(ref[i]) || !std::isfinite(out[i]) || !std::isfinite(ref[i]))
+        {
+            max_err = err > max_err ? err : max_err;
+            err_count++;
+            if(err_count < 5)
+            {
+                std::cout << std::setw(12) << std::setprecision(7) << "out[" << i << "] != ref["
+                          << i << "]: " << out[i] << "!=" << ref[i] << std::endl
+                          << msg << std::endl;
+            }
+            res = false;
+        }
+    }
+    if(!res)
+    {
+        std::cout << std::setw(12) << std::setprecision(7) << "max err: " << max_err << std::endl;
+    }
+    return res;
+}
+
+template <typename T>
+typename std::enable_if<std::is_integral<T>::value, bool>::type check_err(
+    const std::vector<T>& out, const std::vector<T>& ref, const std::string& msg, T = 0, T = 0)
+{
+    if(out.size() != ref.size())
+    {
+        std::cout << "out.size() != ref.size(), :" << out.size() << " != " << ref.size()
+                  << std::endl
+                  << msg << std::endl;
+        return false;
+    }
+
+    for(std::size_t i = 0; i < ref.size(); ++i)
+    {
+        if(out[i] != ref[i])
+        {
+            std::cout << "out[" << i << "] != ref[" << i << "]: " << out[i] << "!=" << ref[i]
+                      << std::endl
+                      << msg << std::endl;
+            return false;
+        }
+    }
+    return true;
+}
+
+} // namespace test_util
+
+#endif