feat(exp): Support get random data from queue

Signed-off-by: KunoiSayami <[email protected]>

1 files changed, 35 insertions, 7 deletions

diff --git a/expt_0510.cu b/expt_0510.cu
index 17e104d..1b95385 100644
--- a/expt_0510.cu
+++ b/expt_0510.cu
@@ -4,6 +4,7 @@
 #include <algorithm>
 #include <cassert>
 #include <cstdio>
+#include <random>
 #include <vector>
 
 std::vector<unsigned long long> population_vector, sample_vector,
@@ -159,16 +160,39 @@ __global__ void freeStorageStage2() {
   // cudaFree(cdf_result);
 }
 
-void read_file(char const *filename, long sample_length,
-               unsigned long max_number = TEST_LENGTH) {
-  max_number += sample_length + 256;
-  auto read_number = 0;
+unsigned long randomRow(unsigned long max_value_) {
+  std::random_device randomDevice;
+  std::mt19937 mt19937(randomDevice());
+  std::uniform_int_distribution<std::mt19937::result_type> dst(0, max_value_);
+  return dst(mt19937);
+}
+
+void readFile(char const *filename, long sample_length,
+              unsigned long &total_row,
+              unsigned long max_number = TEST_LENGTH) {
+  auto read_number = 0UL;
   FILE *file = fopen(filename, "r");
   assert(file);
   for (long long i;
        read_number < max_number && fscanf(file, "%lld ", &i) != EOF;
        store_into_vector(i))
     read_number++;
+
+  if (total_row > 0) {
+    read_number = randomRow(total_row - sample_length - max_number - 256);
+    total_row = read_number;
+    // fprintf(stderr, "Skip %lu\n", read_number);
+    for (long long i; read_number > 0 && fscanf(file, "%lld ", &i) != EOF;
+         store_into_vector(i))
+      read_number--;
+  }
+
+  read_number = 0;
+
+  auto remain = sample_length + 256;
+  for (long long i; read_number < remain && fscanf(file, "%lld ", &i) != EOF;
+       store_into_vector(i))
+    read_number++;
   fclose(file);
 }
 
@@ -188,6 +212,7 @@ int main(int argc, char const *argv[]) {
 
   auto test_size = TEST_LENGTH;
   auto sample_length = DEFAULT_SAMPLE_LENGTH;
+  auto total_row = 0UL;
 
   if (argc >= 2) {
     test_size = strtol(argv[1], nullptr, 10);
@@ -195,11 +220,14 @@ int main(int argc, char const *argv[]) {
   if (argc >= 3) {
     sample_length = pow_for_sample(strtol(argv[2], nullptr, 10));
   }
+  if (argc >= 4) {
+    total_row = strtol(argv[3], nullptr, 10);
+  }
 
-  read_file("normal_distribution.txt", test_size);
+  readFile("normal_distribution.txt", test_size, total_row);
 
-  printf("population: %zu, test size: %zu, sample length: %zu | ",
-         population_vector.size(), test_size, sample_length);
+  printf("population: %zu, skip: %lu, test size: %zu, sample length: %zu | ",
+         population_vector.size(), total_row, test_size, sample_length);
 
   sample_vector = std::vector<key_type>(
       population_vector.begin(), population_vector.begin() + sample_length - 2);