add license in file (#303)

[ROCm/composable_kernel commit: d3051d7517]

library/include/ck/library/host_tensor/conv_common.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/tensor_description/tensor_descriptor.hpp"

library/include/ck/library/host_tensor/device_memory.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <hip/hip_runtime.h>

library/include/ck/library/host_tensor/host_common_util.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <vector>

library/include/ck/library/host_tensor/host_conv.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 #include "host_tensor.hpp"
 #include "conv_common.hpp"

library/include/ck/library/host_tensor/host_gemm.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "host_tensor.hpp"

library/include/ck/library/host_tensor/host_reduction.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <vector>

library/include/ck/library/host_tensor/host_tensor.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <thread>

library/include/ck/library/host_tensor/host_tensor_generator.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <cmath>

library/include/ck/library/reference_tensor_operation/cpu/reference_batched_gemm.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_cgemm.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_conv_backward_weight.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_conv_bwd_data.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_conv_fwd_bias_activation_add.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_gemm.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_2d.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_gemm_bias_activation_add.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/cpu/reference_softmax.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <iostream>

library/include/ck/library/reference_tensor_operation/gpu/naive_conv_fwd.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #ifndef NAIVE_CONV_FWD_HPP
 #define NAIVE_CONV_FWD_HPP
 

library/include/ck/library/tensor_operation_instance/device_operation_instance.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <vector>

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f32_f16.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f32_f32.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f64_f32.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f64_f64_f64.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_i8_i32_i8.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_i8_i8_i8.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_impl_common.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 namespace ck {

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_b16_f32_f32.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_f16_f32_f32.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_f32_f32_f32.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_f32_f64_f32.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_multiblock_atomic_add_f64_f64_f64.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_b16_f32_b16.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f16_f16_f16.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f16_f32_f16.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f32_f32_f32.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f32_f64_f32.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_f64_f64_f64.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_i8_i32_i8.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_threadwise_i8_i8_i8.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include "ck/utility/data_type.hpp"

library/include/ck/library/utility/check_err.hpp

@@ -1,207 +1,210 @@
-#pragma once
-
-#include <algorithm>
-#include <cmath>
-#include <cstdlib>
-#include <iostream>
-#include <iomanip>
-#include <iterator>
-#include <limits>
-#include <type_traits>
-#include <vector>
-
-#include "ck/utility/data_type.hpp"
-
-namespace ck {
-namespace utils {
-
-template <typename T>
-typename std::enable_if<std::is_floating_point<T>::value && !std::is_same<T, half_t>::value,
-                        bool>::type
-check_err(const std::vector<T>& out,
-          const std::vector<T>& ref,
-          const std::string& msg = "Error: Incorrect results!",
-          double rtol            = 1e-5,
-          double atol            = 3e-6)
-{
-    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;
-    double err     = 0;
-    double max_err = std::numeric_limits<double>::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_same<T, bhalf_t>::value, bool>::type
-check_err(const std::vector<T>& out,
-          const std::vector<T>& ref,
-          const std::string& msg = "Error: Incorrect results!",
-          double rtol            = 1e-3,
-          double atol            = 1e-3)
-{
-    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;
-    double err    = 0;
-    // TODO: This is a hack. We should have proper specialization for bhalf_t data type.
-    double max_err = std::numeric_limits<float>::min();
-    for(std::size_t i = 0; i < ref.size(); ++i)
-    {
-        double o = type_convert<float>(out[i]);
-        double r = type_convert<float>(ref[i]);
-        err      = std::abs(o - r);
-        if(err > atol + rtol * std::abs(r) || !std::isfinite(o) || !std::isfinite(r))
-        {
-            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 << "]: " << o << " != " << r << 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_same<T, half_t>::value, bool>::type
-check_err(const std::vector<T>& out,
-          const std::vector<T>& ref,
-          const std::string& msg = "Error: Incorrect results!",
-          double rtol            = 1e-3,
-          double atol            = 1e-3)
-{
-    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;
-    double err     = 0;
-    double max_err = std::numeric_limits<T>::min();
-    for(std::size_t i = 0; i < ref.size(); ++i)
-    {
-        double o = type_convert<float>(out[i]);
-        double r = type_convert<float>(ref[i]);
-        err      = std::abs(o - r);
-        if(err > atol + rtol * std::abs(r) || !std::isfinite(o) || !std::isfinite(r))
-        {
-            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 << "]: " << o << " != " << r << 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 && !std::is_same<T, bhalf_t>::value, bool>::type
-check_err(const std::vector<T>& out,
-          const std::vector<T>& ref,
-          const std::string& msg = "Error: Incorrect results!",
-          double                 = 0,
-          double                 = 0)
-{
-    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;
-    int64_t err     = 0;
-    int64_t max_err = std::numeric_limits<int64_t>::min();
-    for(std::size_t i = 0; i < ref.size(); ++i)
-    {
-        int64_t o = out[i];
-        int64_t r = ref[i];
-        err       = std::abs(o - r);
-
-        if(err > 0)
-        {
-            max_err = err > max_err ? err : max_err;
-            err_count++;
-            if(err_count < 5)
-            {
-                std::cout << "out[" << i << "] != ref[" << i << "]: " << static_cast<int>(out[i])
-                          << " != " << static_cast<int>(ref[i]) << std::endl
-                          << msg << std::endl;
-            }
-            res = false;
-        }
-    }
-    if(!res)
-    {
-        std::cout << "max err: " << max_err << std::endl;
-    }
-    return res;
-}
-
-} // namespace utils
-} // namespace ck
-
-template <typename T>
-std::ostream& operator<<(std::ostream& os, const std::vector<T>& v)
-{
-    std::copy(std::begin(v), std::end(v), std::ostream_iterator<T>(os, " "));
-    return os;
-}
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <algorithm>
+#include <cmath>
+#include <cstdlib>
+#include <iostream>
+#include <iomanip>
+#include <iterator>
+#include <limits>
+#include <type_traits>
+#include <vector>
+
+#include "ck/utility/data_type.hpp"
+
+namespace ck {
+namespace utils {
+
+template <typename T>
+typename std::enable_if<std::is_floating_point<T>::value && !std::is_same<T, half_t>::value,
+                        bool>::type
+check_err(const std::vector<T>& out,
+          const std::vector<T>& ref,
+          const std::string& msg = "Error: Incorrect results!",
+          double rtol            = 1e-5,
+          double atol            = 3e-6)
+{
+    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;
+    double err     = 0;
+    double max_err = std::numeric_limits<double>::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_same<T, bhalf_t>::value, bool>::type
+check_err(const std::vector<T>& out,
+          const std::vector<T>& ref,
+          const std::string& msg = "Error: Incorrect results!",
+          double rtol            = 1e-3,
+          double atol            = 1e-3)
+{
+    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;
+    double err    = 0;
+    // TODO: This is a hack. We should have proper specialization for bhalf_t data type.
+    double max_err = std::numeric_limits<float>::min();
+    for(std::size_t i = 0; i < ref.size(); ++i)
+    {
+        double o = type_convert<float>(out[i]);
+        double r = type_convert<float>(ref[i]);
+        err      = std::abs(o - r);
+        if(err > atol + rtol * std::abs(r) || !std::isfinite(o) || !std::isfinite(r))
+        {
+            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 << "]: " << o << " != " << r << 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_same<T, half_t>::value, bool>::type
+check_err(const std::vector<T>& out,
+          const std::vector<T>& ref,
+          const std::string& msg = "Error: Incorrect results!",
+          double rtol            = 1e-3,
+          double atol            = 1e-3)
+{
+    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;
+    double err     = 0;
+    double max_err = std::numeric_limits<T>::min();
+    for(std::size_t i = 0; i < ref.size(); ++i)
+    {
+        double o = type_convert<float>(out[i]);
+        double r = type_convert<float>(ref[i]);
+        err      = std::abs(o - r);
+        if(err > atol + rtol * std::abs(r) || !std::isfinite(o) || !std::isfinite(r))
+        {
+            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 << "]: " << o << " != " << r << 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 && !std::is_same<T, bhalf_t>::value, bool>::type
+check_err(const std::vector<T>& out,
+          const std::vector<T>& ref,
+          const std::string& msg = "Error: Incorrect results!",
+          double                 = 0,
+          double                 = 0)
+{
+    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;
+    int64_t err     = 0;
+    int64_t max_err = std::numeric_limits<int64_t>::min();
+    for(std::size_t i = 0; i < ref.size(); ++i)
+    {
+        int64_t o = out[i];
+        int64_t r = ref[i];
+        err       = std::abs(o - r);
+
+        if(err > 0)
+        {
+            max_err = err > max_err ? err : max_err;
+            err_count++;
+            if(err_count < 5)
+            {
+                std::cout << "out[" << i << "] != ref[" << i << "]: " << static_cast<int>(out[i])
+                          << " != " << static_cast<int>(ref[i]) << std::endl
+                          << msg << std::endl;
+            }
+            res = false;
+        }
+    }
+    if(!res)
+    {
+        std::cout << "max err: " << max_err << std::endl;
+    }
+    return res;
+}
+
+} // namespace utils
+} // namespace ck
+
+template <typename T>
+std::ostream& operator<<(std::ostream& os, const std::vector<T>& v)
+{
+    std::copy(std::begin(v), std::end(v), std::ostream_iterator<T>(os, " "));
+    return os;
+}

library/include/ck/library/utility/conv_util.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <cstdlib>

library/include/ck/library/utility/fill.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <algorithm>

library/include/ck/library/utility/op_instance_engine.hpp

@@ -1,3 +1,6 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
 #pragma once
 
 #include <cstdlib>