Sakura multiclass hashmap benchmark

sakura_benchmark.R

library(sakura)
library(microbenchmark)
library(dplyr)

class_names <- sapply(1:48, function(i) {
  paste0("class", i)
})

sfuncs <- lapply(1:48, function(i) {
  function(x) as.raw(i)
})

ufuncs <- lapply(1:48, function(i) {
  function(x) {
    if(x != i) stop("wrong serialization function")
    result <- new.env()
    class(result) <- paste0("class",i)
    result
  }
})

config1 <- sakura::serial_config(class = class_names[1],
                                  sfunc = sfuncs[[1]],
                                  ufunc = ufuncs[[1]])

config16 <- sakura::serial_config(class = class_names[1:16],
                                  sfunc = sfuncs[1:16],
                                  ufunc = ufuncs[1:16])

config48 <- sakura::serial_config(class = class_names,
                                  sfunc = sfuncs,
                                  ufunc = ufuncs)

generate_class_list <- function(k, m, n) {
   lapply(1:n, function(i) {
    result <- new.env()
    class_name <- paste0("class", sample.int(k, 1))
    if(m > 0) {
      extra_inherits <- paste0("class", LETTERS[1:m])
      class(result) <- sample(c(class_name, extra_inherits))
    } else {
      class(result) <- class_name
    }
    result
   })
}

list_k1_m0_n1 <- generate_class_list(1,0,1)
list_k1_m0_n20000 <- generate_class_list(1,0,20000)
list_k16_m0_n20000 <- generate_class_list(16,0,20000)
list_k16_m2_n20000 <- generate_class_list(16,2,20000)
list_k48_m2_n20000 <- generate_class_list(48,2,20000)

mb <- microbenchmark(
  k1_m0_n1 = sakura::serialize(list_k1_m0_n1, config1),
  k1_m0_n20000 = sakura::serialize(list_k1_m0_n20000, config1),
  k16_m0_n20000 = sakura::serialize(list_k16_m0_n20000, config16),
  k16_m2_n20000 = sakura::serialize(list_k16_m2_n20000, config16),
  k48_m2_n20000 = sakura::serialize(list_k48_m2_n20000, config48), times = 10, unit = "ms")
print(mb, signif=2)

k1_m0_n1 <- sakura::serialize(list_k1_m0_n1, config1)
k1_m0_n20000 <- sakura::serialize(list_k1_m0_n20000, config1)
k16_m0_n20000 <- sakura::serialize(list_k16_m0_n20000, config16)
k16_m2_n20000 <- sakura::serialize(list_k16_m2_n20000, config16)
k48_m2_n20000 <- sakura::serialize(list_k48_m2_n20000, config48)


mb <- microbenchmark(
  k1_m0_n1 = sakura::unserialize(k1_m0_n1, config1),
  k1_m0_n20000 = sakura::unserialize(k1_m0_n20000, config1),
  k16_m0_n20000 = sakura::unserialize(k16_m0_n20000, config16),
  k16_m2_n20000 = sakura::unserialize(k16_m2_n20000, config16),
  k48_m2_n20000 = sakura::unserialize(k48_m2_n20000, config48), times = 10, unit = "ms")
print(mb, signif=2)

# check result correctness
res <- sakura::unserialize(k48_m2_n20000, config48)
class_original <- sapply(1:length(list_k48_m2_n20000), function(i) {
  class(list_k48_m2_n20000[[i]]) %>% grep("[0-9]", ., value=T)
})
class_unserialized <- sapply(res, class)
all(class_original == class_unserialized) # [1] TRUE