[Assorted-commits] SF.net SVN: assorted: [395] hash-join/trunk/src/hashjoin.cc

[<yan...@us...>]

Revision: 395
          http://assorted.svn.sourceforge.net/assorted/?rev=395&view=rev
Author:   yangzhang
Date:     2008-02-12 13:58:51 -0800 (Tue, 12 Feb 2008)

Log Message:
-----------
tweaks

Modified Paths:
--------------
    hash-join/trunk/src/hashjoin.cc

Modified: hash-join/trunk/src/hashjoin.cc
===================================================================
--- hash-join/trunk/src/hashjoin.cc	2008-02-12 19:16:18 UTC (rev 394)
+++ hash-join/trunk/src/hashjoin.cc	2008-02-12 21:58:51 UTC (rev 395)
@@ -1,3 +1,7 @@
+//
+// Hash Join
+//
+
 #include <memory>
 #include <cassert>
 #include <cstdio>
@@ -23,10 +27,6 @@
 
 // TODO: #include <boost/array.hpp>
 
-//
-// Hash Join
-//
-
 using namespace std;
 using namespace __gnu_cxx;
 using namespace commons;
@@ -46,14 +46,15 @@
   void
   resize(size_type hint)
   {
-    cout << "resizing " << this << " to " << hint << endl;
+    // cout << "resizing " << this << " to " << hint << endl;
     my_hash_map::resize(hint);
   }
 };
 
 // TODO use dependency injection!
 unsigned int ncpus = 1;
-const hmap::size_type map_size = 10000000;
+const hmap::size_type map_size = 1000000;
+const int min_bucket_size = 1000000;
 
 /**
  * Buckets are produced in the hash-partitioning phase. These are simple
@@ -224,7 +225,7 @@
     buckets[i] = new bucket[ncpus];
     for (unsigned int j = 0; j < ncpus; j++) {
       // TODO dependency injection
-      size_t bucket_size = max((size_t) 1000000,buflen / ncpus * 3);
+      size_t bucket_size = max((size_t) min_bucket_size,buflen / ncpus * 3);
       // Each bucket should be twice as large as it would be given uniform
       // distribution. This is just an initial size; extending can happen.
       buckets[i][j].bufs.push_back(new char[bucket_size]);
@@ -260,7 +261,7 @@
 {
   size_t h = hash_djb2(s);
   unsigned int bucket = h % ncpus;
-  size_t bucket_size = max((size_t) 1000000, buflen * 3 / ncpus);
+  size_t bucket_size = max((size_t) min_bucket_size, buflen * 3 / ncpus);
   if (heads[bucket] + nbytes < bs[bucket].bufs.back() + bucket_size) {
     memcpy(heads[bucket], p, nbytes);
     heads[bucket] += nbytes;
@@ -470,7 +471,7 @@
       }
     }
   }
-  cout << "cpu " << pid << " hits " << hits << " misses " << misses << endl;
+  // cout << "cpu " << pid << " hits " << hits << " misses " << misses << endl;
 }
 
 // vim:et:sw=2:ts=2


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.