Merge pull request #4632 from ex-rzr/make-hip-standard-platform

HIP platform

platforms/hip/src/kernels/parallel.hip

+/**
+ * Sum the forces computed by different contexts.
+ */
+
+extern "C" __global__ void sumForces(long long* __restrict__ force, long long* __restrict__ buffer, int bufferSize, int numBuffers) {
+    int totalSize = bufferSize*numBuffers;
+    for (int index = blockDim.x*blockIdx.x+threadIdx.x; index < bufferSize; index += blockDim.x*gridDim.x) {
+        long long sum = force[index];
+        for (int i = index; i < totalSize; i += bufferSize)
+            sum += buffer[i];
+        force[index] = sum;
+    }
+}