Method get_current_tile_for_idx

include/scheduler.h:38–46

    }

    DEVICE TileInfo get_current_tile_for_idx(int64_t idx) {
        int64_t cluster_id = idx / (ClusterX * ClusterY); // ClusterZ should be 1
        auto [block_m_in_cluster, block_n_in_cluster, _] = block_id_in_cluster();
        int cluster_m = cluster_id / (problem_n / ClusterY);
        int cluster_n = cluster_id - cluster_m * (problem_n / ClusterY);
        int work_m = cluster_m * ClusterX + block_m_in_cluster;
        int work_n = cluster_n * ClusterY + block_n_in_cluster; 
        return {work_m, work_n, idx < total_problem_mn};
    }

    DEVICE void advance(int count = 1) {
        current_idx += gridDim.z * gridDim.y * gridDim.x * count;