diff --git a/flang/include/flang/Runtime/CUDA/common.h b/flang/include/flang/Runtime/CUDA/common.h
index b73bc390ea8c9..e9f61932230e9 100644
--- a/flang/include/flang/Runtime/CUDA/common.h
+++ b/flang/include/flang/Runtime/CUDA/common.h
@@ -30,7 +30,7 @@ static constexpr unsigned kDeviceToDevice = 2;
     const char *name = cudaGetErrorName(err); \
     if (!name) \
       name = "<unknown>"; \
-    Terminator terminator{__FILE__, __LINE__}; \
+    Fortran::runtime::Terminator terminator{__FILE__, __LINE__}; \
     terminator.Crash("'%s' failed with '%s'", #expr, name); \
   }(expr)
 
diff --git a/flang/include/flang/Runtime/CUDA/kernel.h b/flang/include/flang/Runtime/CUDA/kernel.h
new file mode 100644
index 0000000000000..cf07d874a082c
--- /dev/null
+++ b/flang/include/flang/Runtime/CUDA/kernel.h
@@ -0,0 +1,27 @@
+//===-- include/flang/Runtime/CUDA/kernel.h ---------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef FORTRAN_RUNTIME_CUDA_KERNEL_H_
+#define FORTRAN_RUNTIME_CUDA_KERNEL_H_
+
+#include "flang/Runtime/entry-names.h"
+#include <cstddef>
+#include <stdint.h>
+
+extern "C" {
+
+// This function uses intptr_t instead of CUDA's unsigned int to match
+// the type of MLIR's index type. This avoids the need for casts in the
+// generated MLIR code.
+void RTDEF(CUFLaunchKernel)(const void *kernelName, intptr_t gridX,
+    intptr_t gridY, intptr_t gridZ, intptr_t blockX, intptr_t blockY,
+    intptr_t blockZ, int32_t smem, void **params, void **extra);
+
+} // extern "C"
+
+#endif // FORTRAN_RUNTIME_CUDA_KERNEL_H_
diff --git a/flang/runtime/CUDA/CMakeLists.txt b/flang/runtime/CUDA/CMakeLists.txt
index 86523b419f871..ce87f3efdc363 100644
--- a/flang/runtime/CUDA/CMakeLists.txt
+++ b/flang/runtime/CUDA/CMakeLists.txt
@@ -17,6 +17,7 @@ add_flang_library(${CUFRT_LIBNAME}
   allocator.cpp
   allocatable.cpp
   descriptor.cpp
+  kernel.cpp
   memory.cpp
   registration.cpp
 )
diff --git a/flang/runtime/CUDA/kernel.cpp b/flang/runtime/CUDA/kernel.cpp
new file mode 100644
index 0000000000000..f81153a1af4bc
--- /dev/null
+++ b/flang/runtime/CUDA/kernel.cpp
@@ -0,0 +1,33 @@
+//===-- runtime/CUDA/kernel.cpp -------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "flang/Runtime/CUDA/kernel.h"
+#include "../terminator.h"
+#include "flang/Runtime/CUDA/common.h"
+
+#include "cuda_runtime.h"
+
+extern "C" {
+
+void RTDEF(CUFLaunchKernel)(const void *kernel, intptr_t gridX, intptr_t gridY,
+    intptr_t gridZ, intptr_t blockX, intptr_t blockY, intptr_t blockZ,
+    int32_t smem, void **params, void **extra) {
+  dim3 gridDim;
+  gridDim.x = gridX;
+  gridDim.y = gridY;
+  gridDim.z = gridZ;
+  dim3 blockDim;
+  blockDim.x = blockX;
+  blockDim.y = blockY;
+  blockDim.z = blockZ;
+  cudaStream_t stream = 0;
+  CUDA_REPORT_IF_ERROR(
+      cudaLaunchKernel(kernel, gridDim, blockDim, params, smem, stream));
+}
+
+} // extern "C"