Hiru Lehian kideak#
forbegizta estandarra: Eskuzko iterazioa (aurre-esleitua).lapply: R-ko estandar funtzional eta sekuentziala.future_lapply: Bertsio paralelizatua.
1. Esperimentua: Zeregin “Merkea”#
Eszenatoki honetan, oso azkar egiten den zerbait egingo dugu: 1.000 zenbakiren batez bestekoa kalkulatu.
n <- 200
data_list <- replicate(n, rnorm(1000), simplify = FALSE)
bench_cheap <- microbenchmark(
for_loop = {
res_for <- vector("list", n)
for (i in 1:n) res_for[[i]] <- mean(data_list[[i]])
},
standard_apply = lapply(data_list, mean),
future_apply = future_lapply(data_list, mean),
times = 10
)
# Taula sortu
kable(summary(bench_cheap), caption = "Zeregin merkearen emaitzak (milisegundoak)")| expr | min | lq | mean | median | uq | max | neval |
|---|---|---|---|---|---|---|---|
| for_loop | 1621.207 | 1649.317 | 1844.2160 | 1738.0125 | 1816.831 | 2798.576 | 10 |
| standard_apply | 570.980 | 572.757 | 679.9542 | 609.3615 | 668.030 | 1070.847 | 10 |
| future_apply | 41958.711 | 44211.130 | 66316.4306 | 45609.9595 | 91943.761 | 131875.262 | 10 |
Zeregin merkearen emaitzak (milisegundoak)
# Irudia sortu
autoplot(bench_cheap) +
labs(title = "Zeregin merkea: Paralelizazioaren gainkarga nabaria da")
2. Esperimentua: Zeregin “Garestia”#
Eszenatoki honetan, lan “astun” bat simulatuko dugu atzerapen txiki bat gehituz (Sys.sleep). Horrek modelizazio estatistiko konplexu bat edo web scraping-a imitatzen du.
n_heavy <- 20
data_heavy <- replicate(n_heavy, rnorm(10), simplify = FALSE)
# Deialdi bakoitzeko 0,1 segundo ematen dituen funtzioa
heavy_func <- function(x) {
Sys.sleep(0.1)
mean(x)
}
bench_expensive <- microbenchmark(
for_loop = {
res_for <- vector("list", n_heavy)
for (i in 1:n_heavy) res_for[[i]] <- heavy_func(data_heavy[[i]])
},
standard_apply = lapply(data_heavy, heavy_func),
future_apply = future_lapply(data_heavy, heavy_func),
times = 2 # Iterazio gutxi, motela baita!
)
# Taula sortu
kable(summary(bench_expensive), caption = "Zeregin garestiaren emaitzak (segundoak)")| expr | min | lq | mean | median | uq | max | neval |
|---|---|---|---|---|---|---|---|
| for_loop | 2014.4471 | 2014.4471 | 2014.9119 | 2014.9119 | 2015.3766 | 2015.3766 | 2 |
| standard_apply | 2009.7203 | 2009.7203 | 2011.6294 | 2011.6294 | 2013.5385 | 2013.5385 | 2 |
| future_apply | 287.4234 | 287.4234 | 298.2098 | 298.2098 | 308.9962 | 308.9962 | 2 |
Zeregin garestiaren emaitzak (segundoak)
# Irudia sortu
autoplot(bench_expensive) +
labs(title = "Zeregin garestia: Future-k alde handiz irabazten du")