/// <summary>
/// An explicitly grouped kernel that uses high-level group extensions.
/// Use the available scan/reduce operations in the namespace ILGPU.Algorithms.ScanReduceOperations.
/// </summary>
static void KernelWithGroupExtensions(ArrayView2D <int, Stride2D.DenseX> data)
{
var globalIndex = Grid.GlobalIndex.X;
// Use the all-reduce algorithm to perform a reduction over all lanes in a warp.
// Every lane in the warp will receive the resulting value.
// Use WarpExtensions.Reduce for faster performance, if you need to have the result
// in the first lane only.
data[globalIndex, 0] = GroupExtensions.AllReduce <int, AddInt32>(1);
// Perform an exclusive scan over all lanes in the whole warp.
data[globalIndex, 1] = GroupExtensions.ExclusiveScan <int, AddInt32>(1);
// Perform an inclusive scan over all lanes in the whole warp.
data[globalIndex, 2] = GroupExtensions.InclusiveScan <int, AddInt32>(1);
// Perform a all reduction using a different reduction logic.
data[globalIndex, 3] = GroupExtensions.AllReduce <int, MinInt32>(Group.IdxX + 1);
}