Since Fedora 33, R is linked against FlexiBLAS, a BLAS/LAPACK wrapper library that enables runtime switching of the optimized backend.
The accompanying R-flexiblas package enables BLAS/LAPACK switching without leaving the R session, as well as setting the number of threads for parallel backends.
$ sudo dnf install R-flexiblas # install FlexiBLAS API interface for R
$ sudo dnf install flexiblas-* # install all available optimized backends
Then, in an R session we see:
library(flexiblas)
# check whether FlexiBLAS is available
flexiblas_avail()
#> [1] TRUE
# get the current backend
flexiblas_current_backend()
#> [1] "OPENBLAS-OPENMP"
# list all available backends
flexiblas_list()
#> [1] "NETLIB" "__FALLBACK__" "BLIS-THREADS" "OPENBLAS-OPENMP"
#> [5] "BLIS-SERIAL" "ATLAS" "OPENBLAS-SERIAL" "OPENBLAS-THREADS"
#> [9] "BLIS-OPENMP"
# set/get the number of threads
flexiblas_set_num_threads(12)
flexiblas_get_num_threads()
#> [1] 12
This is an example of GEMM benchmark for all the backends available. You can run the following code interactively or as a script file.
library(flexiblas)
n <- 2000
runs <- 10
ignore <- "__FALLBACK__"
A <- matrix(runif(n*n), nrow=n)
B <- matrix(runif(n*n), nrow=n)
# load backends
backends <- setdiff(flexiblas_list(), ignore)
idx <- flexiblas_load_backend(backends)
# benchmark
timings <- sapply(idx, function(i) {
flexiblas_switch(i)
# warm-up
C <- A[1:100, 1:100] %*% B[1:100, 1:100]
unname(system.time({
for (j in seq_len(runs))
C <- A %*% B
})[3])
})
results <- data.frame(
backend = backends,
`timing [s]` = timings,
`performance [GFlops]` = (2 * (n / 1000)^3) / timings,
check.names = FALSE)
results[order(results$performance),]
#> backend timing [s] performance [GFlops]
#> 1 NETLIB 56.776 0.2818092
#> 5 ATLAS 5.988 2.6720107
#> 2 BLIS-THREADS 3.442 4.6484602
#> 8 BLIS-OPENMP 3.408 4.6948357
#> 4 BLIS-SERIAL 3.395 4.7128130
#> 6 OPENBLAS-SERIAL 3.206 4.9906425
#> 7 OPENBLAS-THREADS 0.773 20.6985770
#> 3 OPENBLAS-OPENMP 0.761 21.0249671
Authors: Iñaki Úcar, Jarek Prokop, Nick Dirschel