SHOW:
|
|
- or go back to the newest paste.
| 1 | # RAM超過6GB，才可以run!! | |
| 2 | # 否則請更改矩陣大小(mat_size)或是列表的長度(length_list) | |
| 3 | length_list = 30 | |
| 4 | mat_size = c(3000, 3000) | |
| 5 | ||
| 6 | mat = matrix(rnorm(prod(mat_size)), mat_size[1], mat_size[2]) | |
| 7 | mat.list = lapply(1:length_list, function(i) mat) | |
| 8 | ||
| 9 | res1 = function(){
| |
| 10 | mat_sum = matrix(rep(0, prod(mat_size)), mat_size[1], mat_size[2]) | |
| 11 | for (i in 1:length_list){
| |
| 12 | mat_sum = mat_sum + mat.list[[i]] | |
| 13 | } | |
| 14 | mat_sum | |
| 15 | } | |
| 16 | res2 = function() Reduce('+', mat.list)
| |
| 17 | res3 = function() matrix(rowSums(matrix(unlist(mat.list), nrow=prod(mat_size))), nrow=mat_size[1]) | |
| 18 | res4 = function() matrix(colSums(do.call(rbind, lapply(mat.list, as.vector))), nrow=mat_size[1]) | |
| 19 | res5 = function() matrix(rowSums(do.call(cbind, lapply(mat.list, as.vector))), nrow=mat_size[1]) | |
| 20 | library(Rcpp) | |
| 21 | library(RcppArmadillo) | |
| 22 | library(inline) | |
| 23 | sourceCpp(code = ' | |
| 24 | // [[Rcpp::depends(RcppArmadillo)]] | |
| 25 | #include <RcppArmadillo.h> | |
| 26 | #include <stdio.h> | |
| 27 | using namespace Rcpp; | |
| 28 | using namespace arma; | |
| 29 | // [[Rcpp::export]] | |
| 30 | mat list_mat_sum_f(List data_list){
| |
| 31 | int n = data_list.size(); | |
| 32 | SEXP mat1 = data_list[0]; | |
| 33 | NumericMatrix mat1_rmat(mat1); | |
| 34 | mat result_mat(mat1_rmat.begin(), mat1_rmat.nrow(), mat1_rmat.ncol()); | |
| 35 | for(int i = 1; i < n; i++) | |
| 36 | {
| |
| 37 | SEXP tmp_m = data_list[i]; | |
| 38 | NumericMatrix data_m(tmp_m); | |
| 39 | mat working_m(data_m.begin(), data_m.nrow(), data_m.ncol(), false); | |
| 40 | result_mat += working_m; | |
| 41 | } | |
| 42 | return result_mat; | |
| 43 | }') | |
| 44 | res6 = function() list_mat_sum_f(mat.list) | |
| 45 | ||
| 46 | # cmpfun | |
| 47 | library(compiler) | |
| 48 | res1_cmp = cmpfun(res1) | |
| 49 | res2_cmp = cmpfun(res2) | |
| 50 | res3_cmp = cmpfun(res3) | |
| 51 | res4_cmp = cmpfun(res4) | |
| 52 | res5_cmp = cmpfun(res5) | |
| 53 | ||
| 54 | all.equal(tmp <- res1(), res2()) | |
| 55 | # TRUE | |
| 56 | all.equal(tmp, res3()) | |
| 57 | # TRUE | |
| 58 | all.equal(tmp, res4()) | |
| 59 | # TRUE | |
| 60 | all.equal(tmp, res5()) | |
| 61 | # TRUE | |
| 62 | all.equal(tmp, res6()) | |
| 63 | # TRUE | |
| 64 | ||
| 65 | library(rbenchmark) | |
| 66 | benchmark(res1(),res2(),res3(),res4(),res5(),res6(),res1_cmp(),res2_cmp(),res3_cmp(),res4_cmp(),res5_cmp(),replications = 20, columns=c('test', 'replications', 'elapsed','relative', 'user.self'), order='relative')
| |
| 67 | ||
| 68 | # test replications elapsed relative user.self | |
| 69 | # 6 res6() 20 6.49 1.000 6.13 | |
| 70 | # 2 res2() 20 15.58 2.401 8.45 | |
| 71 | # 8 res2_cmp() 20 16.19 2.495 8.61 | |
| 72 | # 1 res1() 20 18.44 2.841 10.61 | |
| 73 | # 7 res1_cmp() 20 19.33 2.978 11.15 | |
| 74 | # 5 res5() 20 56.60 8.721 40.24 | |
| 75 | # 11 res5_cmp() 20 56.79 8.750 40.36 | |
| 76 | # 4 res4() 20 70.48 10.860 54.63 | |
| 77 | # 10 res4_cmp() 20 70.53 10.867 55.17 | |
| 78 | # 3 res3() 20 92.87 14.310 76.21 | |
| 79 | # 9 res3_cmp() 20 93.74 14.444 77.19 |
