removing unnecessary comments

Author: kab163

include/umpire/op/dispatch.hpp

@@ -210,7 +210,6 @@ struct op_caller {
       check_memset_bounds(src, src_record, length);
 #endif
     } else if constexpr (std::is_same_v<Op<resource::host_platform>, device_memset<resource::host_platform>>) {
-      // For device_memset, we expect args to be {value, size}
 #ifdef UMPIRE_ENABLE_BOUNDS_CHECKS
       std::size_t length = get_arg<1>(args...);
       check_memset_bounds(src, src_record, length);
@@ -238,7 +237,6 @@ struct op_caller {
       check_memset_bounds(src, src_record, length);
 #endif
     } else if constexpr (std::is_same_v<Op<resource::host_platform>, device_memset<resource::host_platform>>) {
-      // For device_memset, we expect args to be {value, size}
 #ifdef UMPIRE_ENABLE_BOUNDS_CHECKS
       std::size_t length = get_arg<1>(args...);
       check_memset_bounds(src, src_record, length);

include/umpire/op/hip.cpp

@@ -16,9 +16,6 @@ namespace detail {
  * \param data Pointer to the array
  * \param value The value to set each element to
  * \param count The number of elements to set
- *
- * Example: If data is an int array with 10 elements and value is 5,
- *          this will set all 10 ints to the value 5.
  */
 template <typename T>
 __global__ void umpire_device_memset_kernel(T* data, T value, std::size_t count)
@@ -31,12 +28,6 @@ __global__ void umpire_device_memset_kernel(T* data, T value, std::size_t count)
   }
 }
 
-/**
- * @brief Implementation of device_memset - compiled as HIP code
- *
- * This implementation is in the .cpp file (not header) because it uses
- * hipLaunchKernelGGL, which requires HIP compiler support.
- */
 template <typename T>
 void device_memset(T* ptr, T value, std::size_t count)
 {
@@ -52,7 +43,6 @@ void device_memset(T* ptr, T value, std::size_t count)
     grid_size = max_blocks;
   }
 
-  // Launch kernel to set elements (not bytes)
   hipLaunchKernelGGL(umpire_device_memset_kernel<T>, dim3(grid_size), dim3(block_size), 0, 0,
                      ptr, value, count);
 
@@ -71,7 +61,6 @@ void device_memset(T* ptr, T value, std::size_t count)
 }
 
 // Explicit template instantiations for common types
-// Add more instantiations here as needed for your use cases
 template __global__ void umpire_device_memset_kernel<char>(char*, char, std::size_t);
 template __global__ void umpire_device_memset_kernel<unsigned char>(unsigned char*, unsigned char, std::size_t);
 template __global__ void umpire_device_memset_kernel<short>(short*, short, std::size_t);

include/umpire/op/hip.hpp

@@ -12,7 +12,6 @@
 namespace umpire {
 namespace op {
 
-// Forward declaration of device kernel (defined in hip.cpp with explicit instantiations)
 namespace detail {
 template <typename T>
 __global__ void umpire_device_memset_kernel(T* data, T value, std::size_t count);
@@ -189,12 +188,8 @@ inline void memset(T* ptr, int value, std::size_t count)
  *
  * @tparam T Type of array elements
  * @param ptr Pointer to array
- * @param value Value to set each element to (will be cast to type T)
+ * @param value Value to set each element to
  * @param count Number of elements to set
- *
- * @note Example: device_memset(int_array, 5, 10) sets 10 ints to value 5
- * @note The kernel must be explicitly instantiated in hip.cpp for type T
- * @note Implementation is in hip.cpp (compiled as HIP code)
  */
 template <typename T>
 void device_memset(T* ptr, T value, std::size_t count);
@@ -455,15 +450,10 @@ struct device_memset<resource::hip_platform> {
   /**
    * @brief HIP synchronous device memset using kernel
    *
-   * Sets each element of the array to the specified value using a HIP device kernel.
-   * Unlike standard memset which sets bytes, this sets typed elements.
-   *
    * @tparam T Type of array elements
    * @param ptr Pointer to array
    * @param val Value to set each element to
    * @param len Number of elements to set
-   *
-   * @note Example: exec(int_array, 5, 10) sets 10 ints to value 5
    */
   template <typename T>
   static void exec(T* ptr, T val, std::size_t len) noexcept