Rcpp optimization for double for loop

Question

I am using this double for loop to calculate the value temp. rc and tc are results from two hashmap hash_rr and hash_tt. Is there a way to optimize the double forloop to reduce the computational time?

        rc = as<IntegerMatrix>(hash_rr[std::to_string(r)]);
        tc = as<IntegerMatrix>(hash_tt[std::to_string(t)]);
        

        double temp = 0;
        for (int i = 0; i < rc.nrow(); i++){
          for (int j = 0; j < tc.nrow(); j++){
            a1 = rc(i,0) - (b_eff_prev + 1);
            a2 = rc(i,1);
            b1 = tc(j,0);
            b2 = tc(j,1);
            // Rcpp::Rcout << "a,b: " << NumericVector::create(a1,b1,a2,b2) << std::endl;

            temp += nonstop_grid_prev(a1, b1) * den_cpp(a2, b2, n, p);
          }
        }

I tried parallel computing of the double for loop, there is almost no improvement on the computation time.

omp_set_num_threads(omp_get_max_threads());
#pragma omp parallel for reduction(+:temp)