Merge pull request simongog#83 from simongog/YGWYD_rrr_vector

Address issue simongog#82 for rrr_vector
hmusta · Sep 11, 2013 · 2bc7e2f · 2bc7e2f
2 parents 7196439 + 255f401
commit 2bc7e2f
Show file tree

Hide file tree

Showing 5 changed files with 128 additions and 187 deletions.
diff --git a/benchmark/rrr_vector/src/generate_rnd_bitvector.cpp b/benchmark/rrr_vector/src/generate_rnd_bitvector.cpp
@@ -38,7 +38,7 @@ int main(int argc, char* argv[])
     bit_vector v(length);
     srand(17);
     for (uint64_t i=0; i<v.size(); ++i) {
-        if (rand()%100 < density) {
+        if ((uint64_t)(rand()%100) < density) {
             v[i] = 1;
         }
     }

diff --git a/benchmark/rrr_vector/src/rrr_time_and_space.cpp b/benchmark/rrr_vector/src/rrr_time_and_space.cpp
@@ -7,7 +7,11 @@ using namespace std;
 using namespace sdsl;
 
 #ifndef BLOCK_SIZE
-#define 31 BLOCK_SIZE
+#define BLOCK_SIZE 31
+#endif
+
+#ifndef RANK_SAMPLE_DENS
+#define RANK_SAMPLE_DENS 32
 #endif
 
 using namespace std::chrono;
@@ -44,21 +48,21 @@ uint64_t test_inv_random_access(const t_vec& v, const int_vector<64>& rands, uin
 int main(int argc, char* argv[])
 {
     if (argc < 3) {
-        cout << "Usage: " << argv[0] << " bit_vector_file k" << endl;
-        cout << " generates a rrr_vector<" << BLOCK_SIZE << "> with sample rate k" << endl;
+        cout << "Usage: " << argv[0] << " bit_vector_file" << endl;
+        cout << " generates a rrr_vector<" << BLOCK_SIZE << "> with sample rate " << RANK_SAMPLE_DENS << endl;
         cout << " for the bitvector stored in bit_vector_file and run a benchmark" << endl;
         return 1;
     }
 
-    typedef rrr_vector<BLOCK_SIZE> rrr_vec_type;
+    typedef rrr_vector<BLOCK_SIZE, int_vector<>, RANK_SAMPLE_DENS> rrr_vec_type;
     typedef rrr_vec_type::select_1_type rrr_select_type;
     typedef rrr_vec_type::rank_1_type rrr_rank_type;
     bit_vector bv;
     if (load_from_file(bv, argv[1])) {
         cout << "# plain_size = " << size_in_bytes(bv) << endl;
         uint16_t k = atoi(argv[2]);
         auto start = timer::now();
-        rrr_vec_type rrr_vector(bv, k);
+        rrr_vec_type rrr_vector(bv);
         util::clear(bv);
         rrr_select_type rrr_sel(&rrr_vector);
         rrr_rank_type   rrr_rank(&rrr_vector);
@@ -75,21 +79,25 @@ int main(int argc, char* argv[])
         cout << "# btnr_size = "  << size_in_bytes(rrr_vector.btnr) << endl;
         const uint64_t reps = 10000000;
         uint64_t mask = 0;
+        uint64_t check = 0;
         int_vector<64> rands = util::rnd_positions<int_vector<64>>(20, mask, rrr_vector.size(), 17);
         start = timer::now();
-        test_random_access(rrr_vector, rands, mask, reps);
+        check = test_random_access(rrr_vector, rands, mask, reps);
         stop = timer::now();
         cout << "# access_time = " << duration_cast<nanoseconds>(stop-start).count()/(double)reps << endl;
+        cout << "# access_check = " << check << endl;
         rands = util::rnd_positions<int_vector<64>>(20, mask, rrr_vector.size()+1, 17);
         start = timer::now();
-        test_inv_random_access(rrr_rank, rands, mask, reps);
+        check = test_inv_random_access(rrr_rank, rands, mask, reps);
         stop = timer::now();
         cout << "# rank_time = " << duration_cast<nanoseconds>(stop-start).count()/(double)reps << endl;
+        cout << "# rank_check = " << check << endl;
         rands = util::rnd_positions<int_vector<64>>(20, mask, args, 17);
         for (uint64_t i=0; i<rands.size(); ++i) rands[i] = rands[i]+1;
         stop = timer::now();
-        test_inv_random_access(rrr_sel, rands, mask, reps);
+        check = test_inv_random_access(rrr_sel, rands, mask, reps);
         stop = timer::now();
         cout << "# select_time = " << duration_cast<nanoseconds>(stop-start).count()/(double)reps << endl;
+        cout << "# select_check = " << check << endl;
     }
 }
diff --git a/benchmark/rrr_vector/visualize/rrr.R b/benchmark/rrr_vector/visualize/rrr.R
@@ -49,9 +49,10 @@ plot_rrr_space <- function(data, max_y, title="", yaxis=T, xaxis=T, color){
 plot_rrr_query_times <- function( data, max_y=NA, title="", yaxis=T, xaxis=T){
   cat(title,"\n")
   data <- aggregate(data[c('access_time','rank_time','select_time')], by=c(data['K']), FUN=min)
+  data[c('access_time','rank_time','select_time')] <- data[c('access_time','rank_time','select_time')]/1000.0
   data <- data[order(data[['K']]), ]
 
-  max_runtime <- max( data[['access_time']], data[['rank_time']], data[['select_time']])*1000
+  max_runtime <- max( data[['access_time']], data[['rank_time']], data[['select_time']])
   if ( !is.na(max_y) ){
     max_runtime = max_y
   }
@@ -62,7 +63,7 @@ plot_rrr_query_times <- function( data, max_y=NA, title="", yaxis=T, xaxis=T){
        )
   box("plot", col="grey")    
   axis( 1, at = axTicks(1), labels=xaxis, mgp=c(2,0.5,0), tcl=-0.2, cex.axis=1, las=1 )
-  if ( nr >= xlabnr ){
+  if ( xaxis ){
     xlable <- "Block size K"
     mtext(xlable, side=1, line=2, las=0)
   }
@@ -90,7 +91,7 @@ for ( compile_id in compile_config[["COMPILE_ID"]] ){
   par(mfrow=c(n/2+1,2))
   multi_figure_style( n/2+1, 2 )  
   nr <- 0
-  xlabnr <- 2*(n/2)-2
+  xlabnr <- 2*(n/2)-2+1
   for ( tc_id in tc_config[["TC_ID"]] ){
     dd <- subset(d, d[["TC_ID"]]==tc_id)
     plot_rrr_query_times(dd, 2.1, 
@@ -124,7 +125,7 @@ max_size <- 100*max(d[["rrr_size"]]/d[["plain_size"]])
 multi_figure_style( n/2+1, 2 )  
 
 nr <- 0
-xlabnr <- 2*(n/2)-2
+xlabnr <- 2*(n/2)-2+1
 colors = c('gray80','gray50','gray30')
 for ( tc_id in tc_config[["TC_ID"]] ){
   dd <- subset(d, d[["TC_ID"]]==tc_id)