search

Laurae2 / R_Installation

Fully setting up and installing R in Windows / Debian / Ubuntu (Version: 3.5.1)
3 stars 0 forks source link

R 4.0 (Fri, 20 Mar 2020) #1

Open Laurae2 opened 4 years ago

Laurae2 commented 4 years ago

Latest news: https://developer.r-project.org/blosxom.cgi/R-devel/NEWS

To install R-devel 4.0 from 20 March 2020 or later, install libpcre2-dev with the following:

sudo apt-get install libpcre2-dev

image

Else error: pcre2 library and headers are required

Laurae2 commented 4 years ago

Use the following to install xgboost with R 4.0 and Intel MKL:

git clone --recursive https://github.com/dmlc/xgboost
cd xgboost
mkdir build
cd build
cmake .. -DR_LIB=ON -DUSE_CUDA=ON -DCMAKE_C_COMPILER=/usr/bin/gcc-7 -DCMAKE_CXX_COMPILER=/usr/bin/g++-7 -DUSE_NCCL=ON -DNCCL_ROOT=/usr/lib/x86_64-linux-gnu
make -j 36
make install -I/opt/intel/compilers_and_libraries_2019.4.243/linux/mkl/lib/intel64_lin/
sudo R CMD INSTALL ./R-package

Test case:

library(xgboost)

set.seed(1)
N <- 500000
p <- 100
pp <- 25
X <- matrix(runif(N * p), ncol = p)
betas <- 2 * runif(pp) - 1
sel <- sort(sample(p, pp))
m <- X[, sel] %*% betas - 1 + rnorm(N)
y <- rbinom(N, 1, plogis(m))

tr <- sample.int(N, N * 0.90)

trainer <- function(n_cpus, n_gpus, n_iterations) {

  dtrain <- xgb.DMatrix(X[tr,], label = y[tr])
  dtest <- xgb.DMatrix(X[-tr,], label = y[-tr])
  wl <- list(test = dtest)

  if (n_gpus == 0) {

    pt <- proc.time()
    model <- xgb.train(list(objective = "reg:logistic", eval_metric = "logloss", subsample = 0.8, nthread = n_cpus, eta = 0.10,
                            max_bin = 64, tree_method = "hist"),
                       dtrain, watchlist = wl, nrounds = n_iterations)
    my_time <- proc.time() - pt

  } else {

    pt <- proc.time()
    model <- xgb.train(list(objective = "reg:logistic", eval_metric = "logloss", subsample = 0.8, nthread = n_cpus, eta = 0.10,
                            max_bin = 64, tree_method = "gpu_hist", n_gpus = n_gpus),
                       dtrain, watchlist = wl, nrounds = n_iterations)
    my_time <- proc.time() - pt

  }

  rm(model, dtrain, dtest)
  gc(verbose = FALSE)

  return(my_time)

}

trainer(1, 1, 50)
trainer(1, 0, 50)

Results:

R Under development (unstable) (2020-03-27 r78091) -- "Unsuffered Consequences"
Copyright (C) 2020 The R Foundation for Statistical Computing
Platform: x86_64-pc-linux-gnu (64-bit)

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.

  Natural language support but running in an English locale

R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.

> library(xgboost)
> 
> set.seed(1)
> N <- 500000
> p <- 100
> pp <- 25
> X <- matrix(runif(N * p), ncol = p)
> betas <- 2 * runif(pp) - 1
> sel <- sort(sample(p, pp))
> m <- X[, sel] %*% betas - 1 + rnorm(N)
> y <- rbinom(N, 1, plogis(m))
> 
> tr <- sample.int(N, N * 0.90)
> 
> trainer <- function(n_cpus, n_gpus, n_iterations) {
+   
+   dtrain <- xgb.DMatrix(X[tr,], label = y[tr])
+   dtest <- xgb.DMatrix(X[-tr,], label = y[-tr])
+   wl <- list(test = dtest)
+   
+   if (n_gpus == 0) {
+     
+     pt <- proc.time()
+     model <- xgb.train(list(objective = "reg:logistic", eval_metric = "logloss", subsample = 0.8, nthread = n_cpus, eta = 0.10,
+                             max_bin = 64, tree_method = "hist"),
+                        dtrain, watchlist = wl, nrounds = n_iterations)
+     my_time <- proc.time() - pt
+     
+   } else {
+     
+     pt <- proc.time()
+     model <- xgb.train(list(objective = "reg:logistic", eval_metric = "logloss", subsample = 0.8, nthread = n_cpus, eta = 0.10,
+                             max_bin = 64, tree_method = "gpu_hist", n_gpus = n_gpus),
+                        dtrain, watchlist = wl, nrounds = n_iterations)
+     my_time <- proc.time() - pt
+     
+   }
+   
+   rm(model, dtrain, dtest)
+   gc(verbose = FALSE)
+   
+   return(my_time)
+   
+ }
> 
> trainer(1, 1, 50)
[18:44:17] WARNING: /home/laurae/Downloads/R/xgboost/include/xgboost/generic_parameters.h:36: 
n_gpus: 
    Deprecated. Single process multi-GPU training is no longer supported.
    Please switch to distributed training with one process per GPU.
    This can be done using Dask or Spark.  See documentation for details.
[1] test-logloss:0.687465 
[2] test-logloss:0.682720 
[3] test-logloss:0.678590 
[4] test-logloss:0.675024 
[5] test-logloss:0.671988 
[6] test-logloss:0.669297 
[7] test-logloss:0.666885 
[8] test-logloss:0.664741 
[9] test-logloss:0.662744 
[10]    test-logloss:0.660938 
[11]    test-logloss:0.659223 
[12]    test-logloss:0.657735 
[13]    test-logloss:0.656362 
[14]    test-logloss:0.655067 
[15]    test-logloss:0.653892 
[16]    test-logloss:0.652753 
[17]    test-logloss:0.651746 
[18]    test-logloss:0.650809 
[19]    test-logloss:0.649926 
[20]    test-logloss:0.649096 
[21]    test-logloss:0.648273 
[22]    test-logloss:0.647593 
[23]    test-logloss:0.646822 
[24]    test-logloss:0.646082 
[25]    test-logloss:0.645458 
[26]    test-logloss:0.644867 
[27]    test-logloss:0.644329 
[28]    test-logloss:0.643732 
[29]    test-logloss:0.643188 
[30]    test-logloss:0.642666 
[31]    test-logloss:0.642239 
[32]    test-logloss:0.641737 
[33]    test-logloss:0.641328 
[34]    test-logloss:0.640816 
[35]    test-logloss:0.640435 
[36]    test-logloss:0.640006 
[37]    test-logloss:0.639663 
[38]    test-logloss:0.639272 
[39]    test-logloss:0.638916 
[40]    test-logloss:0.638568 
[41]    test-logloss:0.638268 
[42]    test-logloss:0.637945 
[43]    test-logloss:0.637617 
[44]    test-logloss:0.637404 
[45]    test-logloss:0.637157 
[46]    test-logloss:0.636902 
[47]    test-logloss:0.636716 
[48]    test-logloss:0.636500 
[49]    test-logloss:0.636262 
[50]    test-logloss:0.636045 
   user  system elapsed 
  3.687   0.852   4.666 
> trainer(1, 0, 50)
[1] test-logloss:0.687480 
[2] test-logloss:0.682716 
[3] test-logloss:0.678664 
[4] test-logloss:0.675134 
[5] test-logloss:0.672144 
[6] test-logloss:0.669425 
[7] test-logloss:0.666906 
[8] test-logloss:0.664728 
[9] test-logloss:0.662718 
[10]    test-logloss:0.661065 
[11]    test-logloss:0.659484 
[12]    test-logloss:0.657958 
[13]    test-logloss:0.656504 
[14]    test-logloss:0.655166 
[15]    test-logloss:0.654018 
[16]    test-logloss:0.652931 
[17]    test-logloss:0.651872 
[18]    test-logloss:0.650910 
[19]    test-logloss:0.650024 
[20]    test-logloss:0.649156 
[21]    test-logloss:0.648393 
[22]    test-logloss:0.647713 
[23]    test-logloss:0.646957 
[24]    test-logloss:0.646182 
[25]    test-logloss:0.645562 
[26]    test-logloss:0.644967 
[27]    test-logloss:0.644405 
[28]    test-logloss:0.643904 
[29]    test-logloss:0.643418 
[30]    test-logloss:0.642868 
[31]    test-logloss:0.642303 
[32]    test-logloss:0.641841 
[33]    test-logloss:0.641390 
[34]    test-logloss:0.640920 
[35]    test-logloss:0.640521 
[36]    test-logloss:0.640097 
[37]    test-logloss:0.639677 
[38]    test-logloss:0.639321 
[39]    test-logloss:0.638976 
[40]    test-logloss:0.638593 
[41]    test-logloss:0.638342 
[42]    test-logloss:0.637964 
[43]    test-logloss:0.637667 
[44]    test-logloss:0.637394 
[45]    test-logloss:0.637112 
[46]    test-logloss:0.636879 
[47]    test-logloss:0.636631 
[48]    test-logloss:0.636388 
[49]    test-logloss:0.636186 
[50]    test-logloss:0.635976 
   user  system elapsed 
 19.236   0.167  19.412
Laurae2 commented 4 years ago

Use the following to install LightGBM with R 4.0:

git clone --recursive https://github.com/microsoft/LightGBM
cd LightGBM
nano ./R-package/src/install.libs.R
[change `use_GPU <- FALSE` to `use_GPU <- TRUE`, Ctrl+X + Y + Enter]
sudo Rscript build_r.R

Test case:

library(lightgbm)
library(Matrix)
data(agaricus.train, package = "lightgbm")
train <- agaricus.train
train$data[, 1] <- 1:6513
dtrain <- lgb.Dataset(train$data, label = train$label)
data(agaricus.test, package = "lightgbm")
test <- agaricus.test
dtest <- lgb.Dataset.create.valid(dtrain, test$data, label = test$label)
valids <- list(test = dtest)

params <- list(objective = "regression",
               metric = "rmse",
               device = "gpu",
               gpu_platform_id = 0,
               gpu_device_id = 0,
               nthread = 1,
               boost_from_average = FALSE,
               max_bin = 32)
model <- lgb.train(params,
                   dtrain,
                   2,
                   valids,
                   min_data = 1,
                   learning_rate = 1,
                   early_stopping_rounds = 10)