...

simongog · Jun 7, 2013 · 91749ff · 91749ff
1 parent 8f5de44
commit 91749ff
Show file tree

Hide file tree

Showing 5 changed files with 85 additions and 132 deletions.
diff --git a/CPM2013/generate_figure.sh b/CPM2013/generate_figure.sh
@@ -36,5 +36,5 @@ pdflatex ${tmp_tex_file}
 mv ${tmp_pdf_file} ..
 cd ..
 
-rm ${tmp_dir}/*
+#rm ${tmp_dir}/*
 rm ${basename}_rev
diff --git a/query_performance/query_details.R b/query_performance/query_details.R
@@ -1,24 +1,84 @@
 source("../src/basic_functions.R")
+library("tikzDevice")
 
-fn="web-256MB"
+
+
+
+fn="web-4GB"
 data <- data_frame_from_key_value_pairs( paste("query_performance_details_",fn,".txt", sep="" ) )
 
 data['rtime_full'] <- data['rtime_full']/data['full_queries']
 data['utime_full'] <- data['utime_full']/data['full_queries']
 
+data <- subset(data, data[["pattern_file_name"]]=="../pattern/web-4GB.40.1000.0.75.125.pattern")
+#data <- subset(data, data[["pattern_file_name"]]=="../pattern/web-4GB.40.1000.0.8.12.pattern")
+data <- subset(data, data[["fac_dens"]] < 513)
+
 d_mean <- aggregate(data[c('rtime_full','utime_full','fac_dens','disk_access_per_query','phase')] ,list(data[['fac_dens']],data[['phase']]),mean)
 
 dd <- d_mean[order(d_mean['fac_dens'],d_mean['phase']),]
 phases=6
 tab <- dd[['rtime_full']]
 dim(tab) <- c(phases, length(tab)/phases)
-for(i in seq(2,phases)){
-	tab[i] <- tab[i]-tab[i-1]
+for(i in seq(phases, 2)){
+	tab[i,] <- tab[i,]-tab[i-1,]
 }
 
-mycol=topo.colors(2*phases)[seq(1,2*phases,2)]
-barplot(tab, names.arg=unique(dd[['fac_dens']]), ylab="elapsed time per query phase in [millisec]",col=mycol, xlab="K",
-		main=paste("Detailed rtime of a count query ", "l=40, k=100",sep=""))
-legend("topleft", legend=rev(c("condensed BWT matching","load disk block","build block tree","block tree matching","load text","match pattern")),
-		fill=rev(mycol))
+#mycol <- topo.colors(2*phases)[seq(1,2*phases,2)]
+mycol <- terrain.colors(phases)
+
+fac_denss <- unique(dd[['fac_dens']])
+fac_dens_label <- paste(rep("$",length(fac_denss)), fac_denss,rep("$",length(fac_denss)),sep="")
+
+tikz("/Users/sgog/Downloads/cpm/fig/detailed_query_time.tex", width="2.4", height="2.3")
+
+par( oma=c(0.1,0.1,0.1,0.1) )
+par( mar=c(1.5,1.5,0.1,0.1) )
+
+barplot(tab, names.arg=rep("", length(fac_dens_label)), col=mycol, axes=F)
+
+legend("topleft", legend=rev(c("internal matching","load disk block","build block tree","block tree matching","load text","match pattern")),
+		fill=rev(mycol), bty="o", box.lwd=0, bg="white", inset=c(0.01,-0.05), y.intersp=0.8 )
+
+#grid(col="gray")
+
+barp <- barplot(tab, ylab="Runtime per query phase",col=mycol, add=T, cex.axis=0.8, yaxt="n")
+axis(1, at=barp, seq(1, length(fac_dens_label)), labels=fac_dens_label, cex.axis=0.9, lty=0, line=-0.6)
+axis(2, line=-0.5, cex.axis=0.8, lty=0)
+axis(2, line=-0.2, cex.axis=0.8, labels=F )
+
+dev.off()
+
+
+
+tikz("/Users/sgog/Downloads/cpm/fig/detailed_space.tex", width="2.4", height="2.3")
+
+par( oma=c(0.1,0.1,0.1,0.1) )
+par( mar=c(1.5,1.5,0.1,0.1) )
+
+
+
+data <-  data_frame_from_key_value_pairs("../space_usage/space_web-4GB.txt")
+data <- data[order(data[['fac_dens']]),]
+
+data <- subset(data, data[['fac_dens']] %in% c(0,1,4,16,64,256))
+
+size <- c( data[['compact_text_in_megabyte']], data[['header_in_megabyte']], data[['bp_ct_in_megabyte']], data[['lcp_in_megabyte']], data[['sa_in_megabyte']] )
+
+dim(size) <- c( nrow(data), 5)
+size <-	t(size)
+
+rcol=terrain.colors(5)
+
+barp <- barplot( size/(data[["n"]]/(1024*1024)), col=rcol,
+	     ylab="external index size in percent of orig. text", ylim=c(0, 3.2), yaxt="n"   )
+
+axis(1, at=barp, labels = data[['fac_dens']], cex.axis=0.9, lty=0, line=-0.6)
+axis(2, line=-0.5, cex.axis=0.8, lty=0)
+axis(2, line=-0.2, cex.axis=0.8, labels=F)
+
+legend("topright", legend=rev(c("text representation","header","topology","LCP","text pointer")), fill=rev(rcol),
+	   inset=c(-0.1,-0.05),
+	   box.lwd=0, y.intersp=0.8)
 
+dev.off()
diff --git a/query_performance/query_performance_web-256MB.R b/query_performance/query_performance_web-256MB.R
diff --git a/sdsl/examples/64bit_array2int_vector.cpp b/sdsl/examples/64bit_array2int_vector.cpp
@@ -13,10 +13,6 @@ using namespace sdsl;
 using namespace std;
 
 int main(int argc, char *argv[]){
-	if ( argc < 2 ){
-		cout<<"Usage: "<<argv[0]<<" input_file [output_file] [ouput_bit_width]"<<endl;
-		return 1;
-	}
 	size_t x = util::get_file_size(argv[1]);
 	const int BPI=8;
 	cout<<"file size in bytes = "<<x<<endl;
@@ -33,39 +29,25 @@ int main(int argc, char *argv[]){
 	const size_t BUF_SIZE = 6400000; // BUF_SIZE has to be a multiple of 64
 	uint64_t *buf = (uint64_t*)malloc(BUF_SIZE*BPI);
 	uint64_t max=0;
-	uint8_t width=0;
-	if ( argc < 3 ){
-		size_t frac=0, old_frac=0;
-		for (size_t i=0, len=BUF_SIZE*BPI; i<x; i+=len){
-	//		cout<<i<<endl;
-			frac = 100*i/x;
-			if(frac>old_frac){ cout<<"."; if(frac%10==0) cout<<frac; }
-			old_frac=frac;
-			len = BUF_SIZE*BPI;
-			if ( i+len > x ){
-				len = x-i;
-			}
-			fread((char*)buf, 1, len, f);
-			for (size_t j=0; j<len/BPI; ++j){
-				if ( buf[j] > max )
-					max = buf[j];
-	//			cout<<" "<<buf[j]<<endl;
-			}
+	for (size_t i=0, len=BUF_SIZE*BPI; i<x; i+=len){
+		len = BUF_SIZE*BPI;
+		if ( i+len > x ){
+			len = x-i;
+		}
+		fread((char*)buf, 1, len, f);
+		for (size_t j=0; j<len/BPI; ++j){
+			if ( buf[j] > max )
+				max = buf[j];
+//			cout<<" "<<buf[j]<<endl;
 		}
-		cout<<"Max value: "<<max<<endl;
-		width = bit_magic::l1BP(max)+1;
-	}else{
-		width = atoi(argv[3]);
 	}
-
+	cout<<"Max value: "<<max<<endl;
+	uint8_t width = bit_magic::l1BP(max)+1;
 	cout<<"width="<<(int)width<<endl;
 
 //  (2) scan file, bit-compress values and write to outfile
 	rewind(f); // reset file pointer
 	string ofile = string(argv[1])+".int_vector";
-	if ( argc>2 ){
-		ofile = string(argv[2]);
-	}
 	FILE *of = fopen(ofile.c_str(),"wb"); // open output file
 	if ( of == NULL ){
 		cout<<"ERROR: could not open output file "<<argv[1]<<endl;
@@ -89,12 +71,11 @@ int main(int argc, char *argv[]){
 	free(buf);
 	fclose(f);
 	fclose(of);
+	util::load_from_file(v, ofile.c_str());
+	cout<<"v.size()="<<v.size()<<endl;
+	cout<<"v[0]="<<v[0]<<endl;
 	const bool do_check = false;
 	if ( do_check){
-		util::load_from_file(v, ofile.c_str());
-		cout<<"v.size()="<<v.size()<<endl;
-		cout<<"v[0]="<<v[0]<<endl;
-
 		int_vector<> check;
 		util::load_vector_from_file(check, argv[1], BPI);
 		if ( check.size() != v.size() ){

diff --git a/src/rosa_tikz.hpp b/src/rosa_tikz.hpp
@@ -1,5 +1,5 @@
 #ifndef ROSA_TIKZ
-#define ROSA_TIRKZ
+#define ROSA_TIKZ
 
 #include <vector>
 #include <utility> // for pair