fast way to convert data.table to sparse matrix

library(data.table)
library(Matrix)
library(pipeR)

DT <- data.table(a = c(0,1,0), b = 0:2) %>>% `[`( , lapply(.SD, as.numeric))

sparseMat <- DT %>>%  `[`( , rn := .I) %>>% melt.data.table(id.var = "rn") %>>% `[`(value != 0) %>>%
  `[`( , cn := match(variable, names(DT))) %>>% {
    sparseMatrix(.$rn, .$cn, x = .$value)
  }


# benchmark
library(microbenchmark)
N <- 3e4
p <- 200

microbenchmark(op1 = {matrix(rpois(N*p, 2), N) %>>% data.table %>>% as.matrix %>>% as("CsparseMatrix")},
               op2 = {matrix(rpois(N*p, 2), N) %>>% data.table %>>% (~tmp <- names(.)) %>>%
                   `[`( , rn := .I) %>>% melt.data.table(id.var = "rn") %>>%  `[`(value != 0) %>>%
                   `[`( , cn := match(variable, tmp)) %>>% {
                     sparseMatrix(.$rn, .$cn, x = .$value)
                     }}, times = 20L)
# Unit: milliseconds
#  expr      min       lq     mean   median        uq       max neval
#   op1 312.6052 333.5377 411.8370 445.4457  448.6806  551.1241    20
#   op2 735.5928 910.1411 953.2906 968.4100 1028.0968 1087.8577    20