added new binaries to graphing code

benchmarks/concurrentKNN/octTree/experiments.sh

@@ -1,31 +1,29 @@
 #!/bin/bash
-
+mkdir -p graphs
 #run to quickly test the setup
 python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy,neighbors_bench_lockfree,neighbors_bench_hoh] [36,72,144] [50] [1,10] 3 [3DinCube_20M]
 
 
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 100 [1,10] 3 3DinCube_20M
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 50 [1,10] 3 3DinCube_20M
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 10 [1,10] 3 3DinCube_20M
+# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy,neighbors_bench_lockfree,neighbors_bench_hoh] [1,2,4,8,12,36,72,144] [100] [1] 3 [3DinCube_20M]
+# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] [50] [1,10] 3 3DinCube_20M
+# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] [10] [1,10] 3 3DinCube_20M
+
 
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 10 [1,10] 3 3DinCube_2M
 
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 10 [1,10] 2 2DinCube_20M
+# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] [50] [1,10] 2 [2DinCube_20M]
 
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 10 [1,10] 3 3Dplummer_20M
+# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [144] [50] [1,10] 3 [3DinCube_2M,3DinCube_20M,3DinCube_200M,3DinCube_2B]
 
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 10 [1,10] 3 lucy3D_2M
+# python3 run_experiments.py [neighbors_bench] [144] [0,25,50,75,100] [1] 3 [3Dplummer_20M]
+# python3 run_experiments.py [neighbors_bench] [144] [0,25,50,75,100] [1] 3 [3DinCube_20M]
 
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 10 [1,10] 3 lucy3D_14M
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 50 [1,10] 3 lucy3D_14M
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 100 [1,10] 3 lucy3D_14M
+# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] [10] [1,10] 3 3DPlummer_20M
 
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 10 [1,10] 3 thai_statue2M
+# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] [50] [1,10] 3 lucy3D_14M
 
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 10 [1,10] 3 thai_statue5M
-# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] 50 [1,10] 3 thai_statue5M
+# python3 run_experiments.py [neighbors_bench,neighbors_bench_path_copy] [1,2,4,8,12,36,72,144] [50] [1,10] 3 thai_statue5M
 
-# python3 run_experiments.py [range_bench,range_bench_path_copy] [1,2,4,8,12,36,72,144] 10 [.014,.0176] 3 3Dplummer_20M
+# python3 run_experiments.py [range_bench,range_bench_path_copy] [1,2,4,8,12,36,72,144] [10] [.014,.0176] 3 [3Dplummer_20M]
 
 
 

benchmarks/concurrentKNN/octTree/flock/spin_lock.h

@@ -7,7 +7,7 @@
 //   is_locked() -> bool
 
 #include <parlay/parallel.h> // needed for worker_id
-#include <flock/epoch.h>  // for worker_id()
+#include "epoch.h"  // for worker_id()
 
 #include <atomic>
 #include<chrono>

benchmarks/concurrentKNN/octTree/neighbors_bench.h

@@ -82,6 +82,7 @@ void ANN(parlay::sequence<vtx*> &v, int k, int p, double trial_time, int update_
     std::cout << "update_percent: " << update_percent << std::endl;
     std::cout << "query size: " << k << std::endl;
     std::cout << "trial_time: " << trial_time << std::endl;
+    std::cout << "index_size: " << v.size()/2 << std::endl;
 
     //calculate bounding box around the whole point set
     box whole_box = knn_tree::o_tree::get_box(v);     

benchmarks/concurrentKNN/octTree/run_experiments.py

@@ -30,22 +30,42 @@
 
 data_options = {
   "2DinCube_20M" : "../geometryData/data/2DinCube_20000000",
+  "3DinCube_2K" : "../geometryData/data/3DinCube_2000",
+  "3DinCube_20K" : "../geometryData/data/3DinCube_20000",
+  "3DinCube_200K" : "../geometryData/data/3DinCube_200000",
   "3DinCube_2M" : "../geometryData/data/3DinCube_2000000",
   "3DinCube_20M" : "../geometryData/data/3DinCube_20000000",
+  "3DinCube_200M" : "/ssd0/geometryData/3DinCube_200M",
+  "3DinCube_2B" : "/ssd0/geometryData/3DinCube_2B",
   "3Dplummer_20M" : "../geometryData/data/3Dplummer_20000000",
   "lucy3D_2M" : "/ssd0/angel/lucy3D_2M",
   "lucy3D_14M" : "/ssd0/angel/lucy3D_14M",
   "thai_statue2M" : "/ssd0/thai_statue/thai_statue2M",
   "thai_statue5M" : "/ssd0/thai_statue/thai_statue5M",
 }
 
+data_sizes = {
+  "2DinCube_20M" : 10000000,
+  "3DinCube_2K" : 1000,
+  "3DinCube_20K" : 10000,
+  "3DinCube_200K" : 100000,
+  "3DinCube_2M" : 1000000,
+  "3DinCube_20M" : 10000000,
+  "3DinCube_200M" : 100000000,
+  "3Dplummer_20M" : 10000000,
+  "lucy3D_2M" : 1000000,
+  "lucy3D_14M" : 7000000,
+  "thai_statue2M" : 1000000,
+  "thai_statue5M" : 2500000,
+}
+
 parser = argparse.ArgumentParser()
 parser.add_argument("ds_type", help="datastructure type")
 parser.add_argument("threads", help="Number of threads")
 parser.add_argument("ratios", help="Update ratio, number between 0 and 100.")
 parser.add_argument("query_sizes", help="size of query (k)")
 parser.add_argument("dimension", help="dimension of data")
-parser.add_argument("input_name", help="input name")
+parser.add_argument("input_names", help="input names")
 parser.add_argument("-t", "--test_only", help="test script",
                     action="store_true")
 parser.add_argument("-g", "--graphs_only", help="graphs only",
@@ -55,10 +75,10 @@
 args = parser.parse_args()
 print("datastructure: " + args.ds_type)
 print("threads: " + args.threads)
-print("update percent: " + args.ratios)
+print("update percents: " + args.ratios)
 print("query_sizes: " + args.query_sizes)
 print("dimension: " + args.dimension)
-print("input_name: " + args.input_name)
+print("input_names: " + args.input_names)
 
 test_only = args.test_only
 graphs_only = args.graphs_only
@@ -94,7 +114,7 @@ def runstring(test, op, outfile, k):
 
 def runtest(test,procs,u,k,d,infile,extra,outfile) :
     r = rounds
-    otherargs = " -c -t 10.0 "
+    otherargs = " -c -t 1.0 "
 
     runstring(test, "PARLAY_NUM_THREADS=" + str(min(int(procs), maxcpus)) + " numactl -i all ./" + test + " -r " + str(r) + " -d " + str(d) + " -k " + str(k) + " -p " + str(procs) + extra + " -u " + str(u) + otherargs + " " + infile, outfile, k)
 
@@ -105,8 +125,7 @@ def runtest(test,procs,u,k,d,infile,extra,outfile) :
 ratios = args.ratios
 query_sizes = args.query_sizes
 dimension = args.dimension
-input_name = args.input_name
-input_file = data_options[input_name]
+input_names = args.input_names
 
 if '[' in args.threads:
   exp_type = "scalability"
@@ -127,14 +146,32 @@ def runtest(test,procs,u,k,d,infile,extra,outfile) :
   print('invalid argument')
   exit(1)
 
+if '[' in args.ratios:
+  ratios = string_to_list(ratios)
+else:
+  print('invalid argument')
+  exit(1)
+
+if '[' in args.input_names:
+  input_names = string_to_list(input_names)
+else:
+  print('invalid argument')
+  exit(1)
+
+input_files = []
+for name in input_names:
+  input_files.append(data_options[name])
+print(input_files)
+
+
 
 for i in ds_types:
   if i.find("range") != -1:
     exp_category = "range"
   else:
     exp_category = "neighbors"
 
-outfile = "results/" + exp_category + "-".join([ratios+"up", input_name]) + ".txt"
+outfile = "results/" + exp_category + "-".join([input_names[0]]) + ".txt"
 
 datastructures=[]
 for ds in ds_types:
@@ -143,32 +180,40 @@ def runtest(test,procs,u,k,d,infile,extra,outfile) :
 if not graphs_only:
   # clear output file
   os.system("echo \"\" > " + outfile)
-  for ds in datastructures:
-    for th in to_list(threads):
-      for k in to_list(query_sizes):
-        runtest(ds,th,ratios,k,dimension,input_file,"",outfile)
+  for file in input_files:
+    for ds in datastructures:
+      for th in to_list(threads):
+        for k in to_list(query_sizes):
+          for u in to_list(ratios):
+            runtest(ds,th,u,k,dimension,file,"",outfile)
 
 throughput = {}
 stddev = {}
 threads = []
 ratios = []
 algs = []
+sizes = []
 
-readResultsFile(outfile, throughput, stddev, threads, ratios, algs)
+readResultsFile(outfile, throughput, stddev, threads, ratios, sizes, algs)
 
 threads.sort()
 ratios.sort()
 
 print('threads: ' + str(threads))
 print('update ratios: ' + str(ratios))
 print('algs: ' + str(algs))
-print(throughput)
 
-graph_name=exp_category+input_name
+graph_name=exp_category+input_names[0]
 
 alg_names=[]
 for ds in ds_types:
   for k in query_sizes:
     alg_names.append(ds+"-"+str(k))
 
-plot_scalability_graphs(throughput, stddev, threads, ratios, alg_names, graph_name, args.paper_ver)
+
+if(len(input_names) <= 1):
+  plot_scalability_graphs(throughput, stddev, threads, ratios, sizes[0], alg_names, "scalability_"+graph_name, args.paper_ver)
+  plot_ratio_graphs(throughput, stddev, threads, ratios, sizes[0], alg_names, "ratio_"+graph_name, args.paper_ver)
+else:
+  plot_size_graphs(throughput, stddev, threads, ratios, sizes, alg_names, "sizes_"+graph_name, args.paper_ver)
+

benchmarks/concurrentKNN/octTree/verlib/timestamps.h

@@ -1,7 +1,7 @@
 #pragma once
 #include <limits>
 #include <atomic>
-#include "flock/epoch.h"
+#include "../flock/epoch.h"
 #include <x86intrin.h>
 
 namespace verlib {

benchmarks/concurrentKNN/octTree/verlib/verlib.h

@@ -9,7 +9,7 @@
 #pragma once
 #include <parlay/parallel.h>
 #include <parlay/sequence.h>
-#include "flock/flock.h"
+#include "../flock/flock.h"
 
 namespace verlib {
  bool strict_lock = false;

benchmarks/concurrentKNN/octTree/verlib/versioned_hybrid.h

@@ -1,6 +1,6 @@
 #pragma once
 
-#include "flock/flock.h"
+#include "../flock/flock.h"
 #include "timestamps.h"
 
 namespace verlib {

benchmarks/concurrentKNN/octTree/verlib/versioned_indirect.h

@@ -1,5 +1,5 @@
 #pragma once
-#include "flock/flock.h"
+#include "../flock/flock.h"
 #include "timestamps.h"
 
 namespace verlib {

benchmarks/concurrentKNN/octTree/verlib/versioned_recorded_once.h

@@ -9,7 +9,7 @@
 // PPoPP 2021
 
 #pragma once
-#include "flock/flock.h"
+#include "../flock/flock.h"
 #include "timestamps.h"
 
 namespace verlib {