From 96ece41d4116ba88a75943f37e55eb4c3f0c8cdd Mon Sep 17 00:00:00 2001
From: Stefan Brass <stefan.brass@informatik.uni-halle.de>
Date: Sat, 5 Oct 2019 19:05:41 +0200
Subject: [PATCH] bench_cost.sql was missing in load_data
---
db/create_db.sql | 6 +-
db/load_data | 3 +
graph/bench_cost.sql | 45 +++++++
graph/graph.cpp | 281 +++++++++++++++++++++++++++++++++----------
4 files changed, 268 insertions(+), 67 deletions(-)
create mode 100644 graph/bench_cost.sql
diff --git a/db/create_db.sql b/db/create_db.sql
index 911e01a..38e5508 100644
--- a/db/create_db.sql
+++ b/db/create_db.sql
@@ -2,7 +2,7 @@
-- Project: rbench - Logic Programming and Database Benchmarks
-- Filename: db/create_db.sql
-- Purpose: Database for Benchmark Runs: Create Tables and Views
--- Last Change: 19.03.2019
+-- Last Change: 05.10.2019
-- Language: SQL (Tested with PostgreSQL)
-- Author: Stefan Brass
-- EMail: brass@informatik.uni-halle.de
@@ -533,9 +533,9 @@ CREATE TABLE INPUT_JOIN1 (
CREATE TABLE BENCH_COST (
BENCH VARCHAR(20) NOT NULL,
FILE_ID VARCHAR(10) NOT NULL,
- SIZE NUMERIC(12) NOT NULL,
+ SIZE NUMERIC(18) NOT NULL,
ITER NUMERIC(12) NOT NULL,
- INST NUMERIC(12) NOT NULL,
+ INST NUMERIC(18) NOT NULL,
CONSTRAINT BENCH_COST_PK
PRIMARY KEY(BENCH, FILE_ID),
CONSTRAINT BENCH_COST_REF_BENCHMARK
diff --git a/db/load_data b/db/load_data
index 060d7e8..03c1a47 100755
--- a/db/load_data
+++ b/db/load_data
@@ -95,6 +95,9 @@ psql -f graph_types.sql | grep -v "INSERT 0 1"
psql -f input_graphs.sql | grep -v "INSERT 0 1"
psql -f input_join1.sql | grep -v "INSERT 0 1"
+# Cost measures of a benchmark for an input file:
+psql -f bench_cost.sql | grep -v "INSERT 0 1"
+
# Convert tsv-files with benchmark results to SQL:
# (might be unnecessary, if the SQL files already exist, but is safer,
# because SQL files might be missing or outdated)
diff --git a/graph/bench_cost.sql b/graph/bench_cost.sql
new file mode 100644
index 0000000..686d8e2
--- /dev/null
+++ b/graph/bench_cost.sql
@@ -0,0 +1,45 @@
+INSERT INTO BENCH_COST VALUES('tcff','b17',1966082,16,1966082);
+INSERT INTO BENCH_COST VALUES('tcff','b18',4194306,17,4194306);
+INSERT INTO BENCH_COST VALUES('tcff','c1k',1000000,1000,1001000);
+INSERT INTO BENCH_COST VALUES('tcff','c2k',4000000,2000,4002000);
+INSERT INTO BENCH_COST VALUES('tcff','c3k',9000000,3000,9003000);
+INSERT INTO BENCH_COST VALUES('tcff','c4k',16000000,4000,16004000);
+INSERT INTO BENCH_COST VALUES('tcff','k100',10000,1,1010000);
+INSERT INTO BENCH_COST VALUES('tcff','k1k',1000000,1,1001000000);
+INSERT INTO BENCH_COST VALUES('tcff','k1k5',2250000,1,3377250000);
+INSERT INTO BENCH_COST VALUES('tcff','k2k',4000000,1,8004000000);
+INSERT INTO BENCH_COST VALUES('tcff','k50',2500,1,127500);
+INSERT INTO BENCH_COST VALUES('tcff','k500',250000,1,125250000);
+INSERT INTO BENCH_COST VALUES('tcff','m4_2ki',12288,3,12288);
+INSERT INTO BENCH_COST VALUES('tcff','m16_512',61440,15,61440);
+INSERT INTO BENCH_COST VALUES('tcff','m64_128',258048,63,258048);
+INSERT INTO BENCH_COST VALUES('tcff','m256_32',1044480,255,1044480);
+INSERT INTO BENCH_COST VALUES('tcff','m1ki_8',4190208,1023,4190208);
+INSERT INTO BENCH_COST VALUES('tcff','m4ki_2',16773120,4095,16773120);
+INSERT INTO BENCH_COST VALUES('tcff','p1k',499500,999,499500);
+INSERT INTO BENCH_COST VALUES('tcff','p2k',1999000,1999,1999000);
+INSERT INTO BENCH_COST VALUES('tcff','p3k',4498500,2999,4498500);
+INSERT INTO BENCH_COST VALUES('tcff','p4k',7998000,3999,7998000);
+INSERT INTO BENCH_COST VALUES('tcff','s1k_1',1000000,500,2002000);
+INSERT INTO BENCH_COST VALUES('tcff','s1k_3',1000000,250,4004000);
+INSERT INTO BENCH_COST VALUES('tcff','s1k_4',1000000,200,5005000);
+INSERT INTO BENCH_COST VALUES('tcff','s2k_1',4000000,1000,8004000);
+INSERT INTO BENCH_COST VALUES('tcff','s2k_3',4000000,500,16008000);
+INSERT INTO BENCH_COST VALUES('tcff','s2k_4',4000000,400,20010000);
+INSERT INTO BENCH_COST VALUES('tcff','t50',1225,1,20825);
+INSERT INTO BENCH_COST VALUES('tcff','t100',4950,1,166650);
+INSERT INTO BENCH_COST VALUES('tcff','t500',124750,1,20833250);
+INSERT INTO BENCH_COST VALUES('tcff','t1k',499500,1,166666500);
+INSERT INTO BENCH_COST VALUES('tcff','t1k5',1124250,1,562499750);
+INSERT INTO BENCH_COST VALUES('tcff','t2k',1999000,1,1333333000);
+INSERT INTO BENCH_COST VALUES('tcff','v10',8194,9,8194);
+INSERT INTO BENCH_COST VALUES('tcff','v11',18434,10,18434);
+INSERT INTO BENCH_COST VALUES('tcff','v12',40962,11,40962);
+INSERT INTO BENCH_COST VALUES('tcff','v17',1966082,16,1966082);
+INSERT INTO BENCH_COST VALUES('tcff','v18',4194306,17,4194306);
+INSERT INTO BENCH_COST VALUES('tcff','w1k',1000000,1,1000000);
+INSERT INTO BENCH_COST VALUES('tcff','x10k',100020000,2,100020000);
+INSERT INTO BENCH_COST VALUES('tcff','y1k_4k',11998000,4000,11998000);
+INSERT INTO BENCH_COST VALUES('tcff','y1k_8k',39996000,8000,39996000);
+INSERT INTO BENCH_COST VALUES('tcff','y500_4k',9998000,4000,9998000);
+INSERT INTO BENCH_COST VALUES('tcff','y500_8k',35996000,8000,35996000);
diff --git a/graph/graph.cpp b/graph/graph.cpp
index efd6a0b..ebdfeb9 100644
--- a/graph/graph.cpp
+++ b/graph/graph.cpp
@@ -154,12 +154,32 @@
#define PRED_NAME "par"
//=============================================================================
-// Should Test Output be Written?
+// Should Test/Debug Output be Written?
//=============================================================================
// 0 means no test output, 1 means test output:
#define TEST_OUTPUT 0
+//=============================================================================
+// Generate new version of S-Graph?
+//=============================================================================
+
+// 0 means old version (Mario Wenzel), 1 means new version (Stefan Brass):
+#define S_GRAPH_NEW_VER 0
+
+//=============================================================================
+// Name of file with INSERT statements for benchmark cost information:
+//=============================================================================
+
+#define COST_FILE "bench_cost.sql"
+
+//=============================================================================
+// Name of table with benchmark cost information:
+//=============================================================================
+
+#define COST_TABLE "BENCH_COST"
+
+
//=============================================================================
// Include Files:
//=============================================================================
@@ -167,6 +187,7 @@
#include <iostream>
#include <fstream>
#include <cstdint>
+#include <time.h>
//=============================================================================
// Type for Large Integers (64 Bit):
@@ -543,21 +564,21 @@ class GSize {
// Print defined values:
void print() {
if(num_nodes_ >= 0)
- std::cout << "Nodes: " << num_nodes_ << "\n";
+ std::cout << "\tNodes: " << num_nodes_ << "\n";
if(num_edges_ >= 0)
- std::cout << "Edges: " << num_edges_ << "\n";
+ std::cout << "\tEdges: " << num_edges_ << "\n";
if(tc_iter_ >= 0)
- std::cout << "TC Iter: " << tc_iter_ << "\n";
+ std::cout << "\tTC Iter: " << tc_iter_ << "\n";
if(tc_size_ >= 0)
- std::cout << "TC Size: " << tc_size_ << "\n";
+ std::cout << "\tTC Size: " << tc_size_ << "\n";
if(tc_inst_ >= 0)
- std::cout << "TC Inst: " << tc_inst_ << "\n";
+ std::cout << "\tTC Inst: " << tc_inst_ << "\n";
if(sg_iter_ >= 0)
- std::cout << "SG Iter: " << sg_iter_ << "\n";
+ std::cout << "\tSG Iter: " << sg_iter_ << "\n";
if(sg_size_ >= 0)
- std::cout << "SG Size: " << sg_size_ << "\n";
+ std::cout << "\tSG Size: " << sg_size_ << "\n";
if(sg_inst_ >= 0)
- std::cout << "SG Inst: " << sg_inst_ << "\n";
+ std::cout << "\tSG Inst: " << sg_inst_ << "\n";
}
// Check consistency of two graph size objects:
@@ -1487,8 +1508,11 @@ void gen_graph_s(int n, int k, output_t out, gsize_t gsize) {
for(int j = 1; j <= k; j++) {
int skip = (n*j)/(k+1);
for(int i = 1; i <= n; i++) {
- //out->write_edge(i, 1 + (i-1 + skip) % n);
+#if S_GRAPH_NEW_VER
out->write_edge(i, 1 + (i + skip) % n);
+#else
+ out->write_edge(i, 1 + (i-1 + skip) % n);
+#endif
}
}
@@ -1503,14 +1527,7 @@ void gen_graph_s(int n, int k, output_t out, gsize_t gsize) {
gsize->set_tc_inst(r);
// SG Size:
- //if(k == 1 && n % 4 == 2) {
- // gsize->set_sg_size(m * m / 2);
- // gsize->set_sg_inst(2 * (m * m + m));
- //}
- //else {
- // gsize->set_sg_size(m * m);
- // gsize->set_sg_inst(m * (k + 1) + m * m * (k+1) * (k+1));
- //}
+#if S_GRAPH_NEW_VER
gsize->set_sg_size(m * (k+1));
//gsize->set_sg_inst(m*(k+1) + (k+1) * (m * (k+1)) * (k+1));
//The following is the same, slightly shorter:
@@ -1519,6 +1536,16 @@ void gen_graph_s(int n, int k, output_t out, gsize_t gsize) {
gsize->set_sg_iter(2);
else
gsize->set_sg_iter(1);
+#else
+ if(k == 1 && n % 4 == 2) {
+ gsize->set_sg_size(m * m / 2);
+ gsize->set_sg_inst(2 * (m * m + m));
+ }
+ else {
+ gsize->set_sg_size(m * m);
+ gsize->set_sg_inst(m * (k + 1) + m * m * (k+1) * (k+1));
+ }
+#endif
}
//-----------------------------------------------------------------------------
@@ -1932,22 +1959,69 @@ void gen_graph_a(int n, int k, output_t out, gsize_t gsize) {
int main(int argc, str_t argv[])
{
// Print welcome:
- std::cout << "Welcome to the Graph Data Generator.\n";
- std::cout << "\n";
+ // std::cout << "Welcome to the Graph Data Generator.\n";
+ // std::cout << "\n";
+
+ // Compute elapsed time in seconds:
+ time_t start_time;
+ time(&start_time);
// The program should be called with the graph and the output files:
if(argc < 3) {
- std::cerr << "Usage: ./graph GraphID OutputFile1 ...\n";
+ std::cerr <<
+ "Usage: ./graph [-atsv] GraphID OutputFile1 ...\n";
exit(30);
}
+ // Initialize options:
+ bool opt_tc = false;
+ bool opt_sg = false;
+ bool opt_verbose = false;
+
+ // Parse Options:
+ int argno = 1;
+ while(argno < argc && argv[argno][0] == '-') {
+ str_t opts = argv[argno++];
+ opts++; // Skip '-'
+ while(*opts) {
+ switch(*opts) {
+ case 'a':
+ // Compute all benchmarks:
+ opt_tc = true;
+ opt_sg = true;
+ break;
+ case 's':
+ // Compute same generation:
+ opt_sg = true;
+ break;
+ case 't':
+ // Compute transitive closure:
+ opt_tc = true;
+ break;
+ case 'v':
+ // Verbose output (e.g. show benchmark result):
+ opt_verbose = true;
+ break;
+ default:
+ // Unknown option:
+ std::cerr << "Unknown option: " <<
+ "'" << *opts << "'\n";
+ exit(31);
+ }
+ }
+ }
+
// Get graph parameters, first code:
- str_t graph_id = argv[1];
+ if(argno >= argc) {
+ std::cerr << "Command line argument for Graph ID missing.\n";
+ exit(32);
+ }
+ str_t graph_id = argv[argno++];
str_t p = graph_id;
char graph_code = *p++;
if(graph_code == '\0') {
std::cerr << "Impossible empty Graph ID.\n";
- exit(31);
+ exit(33);
}
// First parameter:
@@ -1958,7 +2032,7 @@ int main(int argc, str_t argv[])
par_chars[i] = 0;
if(i == 0) {
std::cerr << "First parameter in graph ID missing.\n";
- exit(32);
+ exit(34);
}
int par1 = str_int(par_chars);
if(*p == 'k') {
@@ -1993,7 +2067,7 @@ int main(int argc, str_t argv[])
par_chars[i] = 0;
if(i == 0) {
std::cerr << "Second parameter in graph ID missing.\n";
- exit(33);
+ exit(35);
}
par2 = str_int(par_chars);
if(*p == 'k') {
@@ -2023,13 +2097,16 @@ int main(int argc, str_t argv[])
// Check that we have successfully parsed the entire graph ID:
if(*p != '\0') {
std::cerr << "Unexpected characters at the end of graph ID.\n";
- exit(34);
+ exit(36);
}
+ // Some feedback about graph to be generated:
+ std::cout << "Generating graph " << graph_id << " ...\n";
+
// Open output file(s):
Output output;
- for(int arg_no = 2; arg_no < argc; arg_no++)
- output.add_file(argv[arg_no]);
+ while(argno < argc)
+ output.add_file(argv[argno++]);
// Create object for graph/benchmark size (computed with formula):
GSize formula;
@@ -2113,56 +2190,132 @@ int main(int argc, str_t argv[])
}
// Output number of edges written:
- std::cout << "Number of edges written: " << output.num_edges() << ".\n";
+ std::cout << " Number of edges written: " <<
+ output.num_edges() << ".\n";
// Close output file (in case we interrupt computation):
output.close_files();
+ // Create object for graph/benchmark size computed from graph:
+ GSize gsize;
+
// Compute transitive closure data:
- std::cout << "Compute transitive closure data ...\n";
- TC tc(output.graph());
- std::cout << "done\n";
- std::cout << "\n";
+ if(opt_tc) {
+ std::cout << "Compute transitive closure data ...\n";
+ TC tc(output.graph());
+ std::cout << "done\n";
+ std::cout << "\n";
- // Test output:
-#if TEST_OUTPUT
- tc.print();
-#endif
+ // Show computation result:
+ if(opt_verbose)
+ tc.print();
+
+ // Store computed size data in graph size object:
+ tc.export_gsize(&gsize);
+ }
// Compute same generation data:
- std::cout << "Compute same generation data ...\n";
- SG sg(output.graph());
- std::cout << "done\n";
- std::cout << "\n";
+ if(opt_sg) {
+ std::cout << "Compute same generation data ...\n";
+ SG sg(output.graph());
+ std::cout << "done\n";
+ std::cout << "\n";
- // Test output:
-#if TEST_OUTPUT
- sg.print();
-#endif
+ // Show computation result:
+ if(opt_verbose)
+ sg.print();
- // Create object for graph/benchmark size computed from graph:
- GSize gsize;
- tc.export_gsize(&gsize);
- sg.export_gsize(&gsize);
-
- // Print computed size data:
- std::cout << "\n";
- std::cout << "Size Data Computed From Graph:\n";
- std::cout << "=============================\n";
- gsize.print();
- std::cout << "\n";
+ // Store computed size data in graph size object:
+ sg.export_gsize(&gsize);
+ }
// Print size data from formula:
- std::cout << "Size Data Computed With Formulas:\n";
- std::cout << "================================\n";
+ std::cout << " Size Data Computed With Formulas:\n";
+ //std::cout << " ================================\n";
formula.print();
std::cout << "\n";
- // Check consistency:
- if(gsize.consistent(&formula))
- std::cout << "Values are consistent.\n";
- else
- std::cout << "*** VALUES DIFFER ***\n";
- std::cout << "\n";
+ // If size data was computed from graph, print that and check result:
+ if(opt_tc || opt_sg) {
+ std::cout << "\n";
+ std::cout << " Size Data Computed From Graph:\n";
+ //std::cout << " =============================\n";
+ gsize.print();
+ std::cout << "\n";
+
+ // Check consistency:
+ if(gsize.consistent(&formula))
+ std::cout << " Values are consistent.\n";
+ else {
+ std::cerr << " *** SIZE VALUES DIFFER ***\n";
+ exit(36);
+ }
+ std::cout << "\n";
+ }
+
+ // Write cost data. First open output file in append mode:
+ std::ofstream cost_file;
+ cost_file.open(COST_FILE, std::ios::out|std::ios::app);
+ if(cost_file.fail() || !cost_file.is_open()) {
+ std::cerr << "Opening '" << COST_FILE << "' failed.\n";
+ exit(37);
+ }
+
+ // Get tc_size where defined (from formula or computed from graph):
+ bigint_t tc_size = formula.tc_size();
+ if(tc_size < 0)
+ tc_size = gsize.tc_size();
+
+ // Get tc_iter where defined:
+ int tc_iter = formula.tc_iter();
+ if(tc_iter < 0)
+ tc_iter = gsize.tc_iter();
+
+ // Get tc_inst where defined:
+ bigint_t tc_inst = formula.tc_inst();
+ if(tc_inst < 0)
+ tc_inst = gsize.tc_inst();
+
+ // If all three are defined, write cost information:
+ if(tc_size >= 0 && tc_iter >= 0 && tc_inst >= 0) {
+ cost_file << "INSERT INTO " << COST_TABLE << " VALUES(" <<
+ "'tcff'" << "," << "'" << graph_id << "'" << "," <<
+ tc_size << "," << tc_iter << "," << tc_inst << ");\n";
+ }
+
+ // Get sg_size where defined (from formula or computed from graph):
+ bigint_t sg_size = formula.sg_size();
+ if(sg_size < 0)
+ sg_size = gsize.sg_size();
+
+ // Get sg_iter where defined:
+ int sg_iter = formula.sg_iter();
+ if(sg_iter < 0)
+ sg_iter = gsize.sg_iter();
+
+ // Get sg_inst where defined:
+ bigint_t sg_inst = formula.sg_inst();
+ if(sg_inst < 0)
+ sg_inst = gsize.sg_inst();
+
+ // If all three are defined, write cost information:
+ if(sg_size >= 0 && sg_iter >= 0 && sg_inst >= 0) {
+ cost_file << "INSERT INTO " << COST_TABLE << " VALUES(" <<
+ "'sgff'" << "," << "'" << graph_id << "'" << "," <<
+ sg_size << "," << sg_iter << "," << sg_inst << ");\n";
+ }
+
+ // Close file:
+ cost_file.close();
+ if(cost_file.fail()) {
+ std::cerr << "Closing '" << COST_FILE << "' failed.\n";
+ exit(38);
+ }
+
+ // Print elapsed time in seconds:
+ time_t end_time;
+ time(&end_time);
+ int seconds = (int) difftime(end_time, start_time);
+ std::cout << " Elapsed time: " << seconds << "s\n";
}
--
GitLab