assorted-commits

[Assorted-commits] SF.net SVN: assorted: [527] simple-build/trunk

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

Revision: 527
          http://assorted.svn.sourceforge.net/assorted/?rev=527&view=rev
Author:   yangzhang
Date:     2008-02-28 19:59:00 -0800 (Thu, 28 Feb 2008)

Log Message:
-----------
added web publishing script

Modified Paths:
--------------
    simple-build/trunk/README
    simple-build/trunk/src/build-templates/cpp

Added Paths:
-----------
    simple-build/trunk/publish.bash

Modified: simple-build/trunk/README
===================================================================
--- simple-build/trunk/README	2008-02-29 03:37:17 UTC (rev 526)
+++ simple-build/trunk/README	2008-02-29 03:59:00 UTC (rev 527)
@@ -111,9 +111,12 @@
   is to one day be able to generate `autotools` inputs (as well as other kinds
   of software packaging).
 
-- [Boost.Build]: A system for building C++ projects. Documentation hints that
-  it is capable of discovering transitive dependencies as well. This seems like
-  the closest thing to what I wanted.
+- [Boost.Build]: A system for building C++ projects. Can discover only
+  transitive dependencies that can be found without relying on implicit
+  locations (such as those that are found via gcc's `CPATH`, `C_INCLUDE_PATH`,
+  `CXX_INCLUDE_PATH`, etc.). Contains a scripting language that I'm still
+  learning, but for simple configs, the language is terse and appears nice. The
+  community doesn't seem to be as active as CMake's.
 
 - [GNU `make`]: Currently, SimpleBuild targets only `make`.
 
@@ -139,9 +142,9 @@
   database came from here.
 
 - [CMake]: Cross-platform `make`. Uses its own quirky scripting language and
-  has quite a few features. I should probably target this. ["Why the KDE
-  project switched to CMake"] is a nice article on CMake (vs. auto* and SCons).
-  I've been using CMake as well.
+  has quite a few features. Kind of ugly. As always, long-tail configurations
+  are more painful. ["Why the KDE project switched to CMake"] is a nice article
+  on CMake (vs. auto* and SCons). I should target this.
 
 - [WAF]: An even simpler/more barren build system. You actually write
   imperative scripts for each build you need. A bit scary.

Added: simple-build/trunk/publish.bash
===================================================================
--- simple-build/trunk/publish.bash	                        (rev 0)
+++ simple-build/trunk/publish.bash	2008-02-29 03:59:00 UTC (rev 527)
@@ -0,0 +1,7 @@
+#!/usr/bin/env bash
+
+clean=true
+project=simple-build
+websrcs=( README )
+webfiles=()
+. assorted.bash || exit 1


Property changes on: simple-build/trunk/publish.bash
___________________________________________________________________
Name: svn:executable
   + *

Modified: simple-build/trunk/src/build-templates/cpp
===================================================================
--- simple-build/trunk/src/build-templates/cpp	2008-02-29 03:37:17 UTC (rev 526)
+++ simple-build/trunk/src/build-templates/cpp	2008-02-29 03:59:00 UTC (rev 527)
@@ -1,6 +1,7 @@
 OBJDIR   := $(OUTDIR)/$(TARGET)-obj
-CFLAGS   := $(CFLAGS) $(FLAGS)
-CXXFLAGS := $(CXXFLAGS) $(FLAGS)
+FLAGS    := -Wall -Werror $(FLAGS)
+CFLAGS   := $(FLAGS) $(CFLAGS)
+CXXFLAGS := $(FLAGS) $(CXXFLAGS)
 CPPFLAGS := $(CPPFLAGS)
 LDFLAGS  := $(LDFLAGS) $(LIBS) $(AUTOLIBS)
 EXE      := $(OUTDIR)/$(TARGET)


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

[Assorted-commits] SF.net SVN: assorted: [526] sandbox/trunk/src/bash/empty-array.bash

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

Revision: 526
          http://assorted.svn.sourceforge.net/assorted/?rev=526&view=rev
Author:   yangzhang
Date:     2008-02-28 19:37:17 -0800 (Thu, 28 Feb 2008)

Log Message:
-----------
tweak

Modified Paths:
--------------
    sandbox/trunk/src/bash/empty-array.bash

Modified: sandbox/trunk/src/bash/empty-array.bash
===================================================================
--- sandbox/trunk/src/bash/empty-array.bash	2008-02-29 03:35:26 UTC (rev 525)
+++ sandbox/trunk/src/bash/empty-array.bash	2008-02-29 03:37:17 UTC (rev 526)
@@ -6,6 +6,10 @@
 arr=1
 echo "${arr[@]}"
 
+# This works.
+arr=()
+echo "${#arr}"
+
 # This fails; apparently empty arrays expanded this way are unset variables.
 arr=()
 echo "${arr[@]}"


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

[Assorted-commits] SF.net SVN: assorted: [525] sandbox/trunk/src/bash/empty-array.bash

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

Revision: 525
          http://assorted.svn.sourceforge.net/assorted/?rev=525&view=rev
Author:   yangzhang
Date:     2008-02-28 19:35:26 -0800 (Thu, 28 Feb 2008)

Log Message:
-----------
added empty-array test

Added Paths:
-----------
    sandbox/trunk/src/bash/empty-array.bash

Added: sandbox/trunk/src/bash/empty-array.bash
===================================================================
--- sandbox/trunk/src/bash/empty-array.bash	                        (rev 0)
+++ sandbox/trunk/src/bash/empty-array.bash	2008-02-29 03:35:26 UTC (rev 525)
@@ -0,0 +1,11 @@
+#!/usr/bin/env bash
+
+set -o nounset -o errexit
+
+# This works.
+arr=1
+echo "${arr[@]}"
+
+# This fails; apparently empty arrays expanded this way are unset variables.
+arr=()
+echo "${arr[@]}"


Property changes on: sandbox/trunk/src/bash/empty-array.bash
___________________________________________________________________
Name: svn:executable
   + *


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

[Assorted-commits] SF.net SVN: assorted: [524] shell-tools/trunk/src/bash-commons/common.bash

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

Revision: 524
          http://assorted.svn.sourceforge.net/assorted/?rev=524&view=rev
Author:   yangzhang
Date:     2008-02-28 19:31:47 -0800 (Thu, 28 Feb 2008)

Log Message:
-----------
added named, parent-dir (for assorted.bash)

Modified Paths:
--------------
    shell-tools/trunk/src/bash-commons/common.bash

Modified: shell-tools/trunk/src/bash-commons/common.bash
===================================================================
--- shell-tools/trunk/src/bash-commons/common.bash	2008-02-29 03:31:33 UTC (rev 523)
+++ shell-tools/trunk/src/bash-commons/common.bash	2008-02-29 03:31:47 UTC (rev 524)
@@ -350,6 +350,24 @@
   fi
 }
 
+# For use with parent-dir.
+named() {
+  [[ "$( basename "$( realpath "$2" )" )" == "$1" ]] && return 0 || return 1
+}
+
+# Recursively search parent directories until a directory is found that
+# satisfies the given predicate. The predicate is the command formed from the
+# arguments passed in.
+parent-dir() {
+  local dir=.
+  while true ; do
+    if "$@" "$dir" ; then break ; fi
+    if [[ "$( realpath "$dir" )" == / ]] ; then return 1 ; fi
+    dir="$dir/.."
+  done
+  echo "$dir"
+}
+
 is-empty() {
   [[ -f "$1" && "$( stat -c%s "$1" )" == 0 ]]
 }


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

[Assorted-commits] SF.net SVN: assorted: [523] shell-tools/trunk/src/bash-commons/bashrc.bash

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

Revision: 523
          http://assorted.svn.sourceforge.net/assorted/?rev=523&view=rev
Author:   yangzhang
Date:     2008-02-28 19:31:33 -0800 (Thu, 28 Feb 2008)

Log Message:
-----------
added boost build path

Modified Paths:
--------------
    shell-tools/trunk/src/bash-commons/bashrc.bash

Modified: shell-tools/trunk/src/bash-commons/bashrc.bash
===================================================================
--- shell-tools/trunk/src/bash-commons/bashrc.bash	2008-02-29 03:31:18 UTC (rev 522)
+++ shell-tools/trunk/src/bash-commons/bashrc.bash	2008-02-29 03:31:33 UTC (rev 523)
@@ -166,6 +166,7 @@
 #[ -z $OCAMLLIB ] && export OCAMLLIB="/usr/lib/ocaml"
 export CLASSPATH="."
 export TEXINPUTS=":/$HOME/.lyx/layouts/packages//"
+export BOOST_BUILD_PATH=/opt/boost-build/share/boost-build
 
 # unison
 


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

[Assorted-commits] SF.net SVN: assorted: [522] shell-tools/trunk/src/bash-commons/assorted. bash

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

Revision: 522
          http://assorted.svn.sourceforge.net/assorted/?rev=522&view=rev
Author:   yangzhang
Date:     2008-02-28 19:31:18 -0800 (Thu, 28 Feb 2008)

Log Message:
-----------
added assorted.bash

Added Paths:
-----------
    shell-tools/trunk/src/bash-commons/assorted.bash

Added: shell-tools/trunk/src/bash-commons/assorted.bash
===================================================================
--- shell-tools/trunk/src/bash-commons/assorted.bash	                        (rev 0)
+++ shell-tools/trunk/src/bash-commons/assorted.bash	2008-02-29 03:31:18 UTC (rev 522)
@@ -0,0 +1,36 @@
+#!/usr/bin/env bash
+
+# This is a common include file for all project configuration scripts for
+# Assorted Projects.
+
+. common.bash || exit 1
+
+root_site_dir="$( parent-dir named assorted )/assorted-site/trunk"
+clean=
+prof=shell-sf
+base_web_dir=assorted/htdocs
+webdir=$base_web_dir/public_html
+stagedir="$( mktemp -d )"
+
+cp -r "${webfiles[@]}" "$stagedir"
+
+for src in "${websrcs[@]}" ; do
+  if [[ src == README ]]
+  then out=index.html
+  else out="$( basename "${src%.*}.html" )"
+  fi
+  pandoc -s -S --tab-stop=2 \
+      -c http://assorted.sf.net/main.css \
+      -H "$(root_site_dir)/header.html" \
+      -A "$(root_site_dir)/footer.html" \
+      -o "$stagedir/$out" "$src"
+done
+
+tar czf - -C "$stagedir" . |
+ssh $prof "
+  set -o errexit -o nounset
+  if [[ $clean ]] ; then rm -r $webdir ; fi
+  mkdir -p $webdir
+  cd $webdir
+  tar xzf -
+"


Property changes on: shell-tools/trunk/src/bash-commons/assorted.bash
___________________________________________________________________
Name: svn:executable
   + *


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

[Assorted-commits] SF.net SVN: assorted: [521] shell-tools/trunk/src/refresh-links.bash

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

Revision: 521
          http://assorted.svn.sourceforge.net/assorted/?rev=521&view=rev
Author:   yangzhang
Date:     2008-02-28 19:29:14 -0800 (Thu, 28 Feb 2008)

Log Message:
-----------
changed local -> .local; i should really remove this script

Modified Paths:
--------------
    shell-tools/trunk/src/refresh-links.bash

Modified: shell-tools/trunk/src/refresh-links.bash
===================================================================
--- shell-tools/trunk/src/refresh-links.bash	2008-02-29 03:28:55 UTC (rev 520)
+++ shell-tools/trunk/src/refresh-links.bash	2008-02-29 03:29:14 UTC (rev 521)
@@ -6,8 +6,8 @@
 do_force=
 do_time=
 do_verbose=
-crawl_dirs="$HOME/local"
-link_dir="$HOME/local"
+crawl_dirs="$HOME/.local"
+link_dir="$HOME/.local"
 usage="Usage: $cmd_name [OPTIONS] [DIR...]
 If no dirs are specified, then defaults to ${crawl_dirs[@]}
 Options:
@@ -50,7 +50,7 @@
 fi
 
 # Warning: don't change this to remove $link_dir! This is frequently the same
-# as $crawl_dir, e.g. ~/local.
+# as $crawl_dir, e.g. ~/.local.
 for dir in bin include info lib man/man{1,2,3,4,5,6,7,8} sbin ; do
     noerr rm -r "$link_dir/$dir/" || true
 done


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

[Assorted-commits] SF.net SVN: assorted: [520] shell-tools/trunk/src/bash-commons/simple-setup .bash

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

Revision: 520
          http://assorted.svn.sourceforge.net/assorted/?rev=520&view=rev
Author:   yangzhang
Date:     2008-02-28 19:28:55 -0800 (Thu, 28 Feb 2008)

Log Message:
-----------
fixed some unbound variable bugs

Modified Paths:
--------------
    shell-tools/trunk/src/bash-commons/simple-setup.bash

Modified: shell-tools/trunk/src/bash-commons/simple-setup.bash
===================================================================
--- shell-tools/trunk/src/bash-commons/simple-setup.bash	2008-02-27 04:11:18 UTC (rev 519)
+++ shell-tools/trunk/src/bash-commons/simple-setup.bash	2008-02-29 03:28:55 UTC (rev 520)
@@ -5,6 +5,7 @@
 
 new_version=$(( BASH_VERSINFO[0] >= 3 && BASH_VERSINFO[1] >= 1 ))
 host=
+do_remote=
 do_remove=
 do_dev=
 prefix="/opt/${pkg:?}"
@@ -58,7 +59,7 @@
 }
 
 install_raw() {
-    local do_strip_ext="$1" do_copy="$2" dst="$3"
+    local do_strip_ext="$1" do_copy="$2" dst="$3" do_dest_path=
     if [[ "${dst#/}" == "$dst" ]] ; then
         dst="$prefix/$dst"
     else
@@ -80,7 +81,7 @@
         fi
         local dstdir="$( dirname "$dst" )/"
         local dstname="$( basename "$dst" )"
-        local do_dest_path=1
+        do_dest_path=1
     else
         local dstdir="$dst"
 


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

[Assorted-commits] SF.net SVN: assorted: [519] sandbox/trunk/src/py/whoisit.py

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

Revision: 519
          http://assorted.svn.sourceforge.net/assorted/?rev=519&view=rev
Author:   yangzhang
Date:     2008-02-26 20:11:18 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
added simple who-is-it script

Added Paths:
-----------
    sandbox/trunk/src/py/whoisit.py

Added: sandbox/trunk/src/py/whoisit.py
===================================================================
--- sandbox/trunk/src/py/whoisit.py	                        (rev 0)
+++ sandbox/trunk/src/py/whoisit.py	2008-02-27 04:11:18 UTC (rev 519)
@@ -0,0 +1,8 @@
+#!/usr/bin/env python
+
+from socket import *
+s=socket()
+s.bind(('harvard.csail.mit.edu',9876))
+s.listen(1)
+t,u=s.accept()
+print t.getpeername()


Property changes on: sandbox/trunk/src/py/whoisit.py
___________________________________________________________________
Name: svn:executable
   + *


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

[Assorted-commits] SF.net SVN: assorted: [518] configs/trunk

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

Revision: 518
          http://assorted.svn.sourceforge.net/assorted/?rev=518&view=rev
Author:   yangzhang
Date:     2008-02-26 20:00:45 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
added topcoder stuff

Modified Paths:
--------------
    configs/trunk/TODO

Added Paths:
-----------
    configs/trunk/src/contestapplet.conf

Modified: configs/trunk/TODO
===================================================================
--- configs/trunk/TODO	2008-02-26 20:49:46 UTC (rev 517)
+++ configs/trunk/TODO	2008-02-27 04:00:45 UTC (rev 518)
@@ -1 +1,2 @@
 mv bashrc.bash into here from shell-tools
+deal with topcoder

Added: configs/trunk/src/contestapplet.conf
===================================================================
--- configs/trunk/src/contestapplet.conf	                        (rev 0)
+++ configs/trunk/src/contestapplet.conf	2008-02-27 04:00:45 UTC (rev 518)
@@ -0,0 +1,225 @@
+// vim:noet:sw=2:ts=2
+
+// TODO
+// - strip out all comments
+// - trim the unused includes/macros/typedefs/code
+
+// $BEGINREUSE$
+#include <algorithm>
+#include <cassert>
+#include <cctype>
+#include <cmath>
+#include <cstdio>
+#include <cstdlib>
+#include <deque>
+#include <functional>
+#include <iostream>
+#include <map>
+#include <queue>
+#include <set>
+#include <sstream>
+#include <stack>
+#include <string>
+#include <vector>
+using namespace std;
+
+#define dd(x) cout << #x << " = " << x << endl
+#define len length()
+#define sz size()
+#define pb(x) push_back((x));
+#define all(A) (A).begin(), (A).end()
+#define rall(A) (A).rbegin(), (A).rend()
+#define forea(it,xs) for (typeof(xs.begin()) it = xs.begin(); it != xs.end(); it++)
+#define FOR(i,xs) for (typeof((xs).size()) i = 0; i < (xs).size(); i++)
+#define FOR(i,l,h) for (typeof(l) i = (l); i < (h); i++)
+
+// CHOOSE
+//int dx[] = {1,0,-1,0}, dy[] = {0,1,0,-1};
+//int dx[] = {1,1,1,0,0,-1,-1,-1}, dy[] = {1,0,-1,1,-1,1,0,-1};
+
+typedef vector<int> vi;
+typedef vector<string> vs;
+typedef vector<bool> vb;
+typedef vector<vi> vvi;
+typedef vector<vs> vvs;
+typedef vector<vb> vvb;
+typedef long long ll;
+typedef vector vec; // XXX
+typedef string str;
+
+template<typename T> inline T mod(T a, T b) { return (a % b + b) % b; }
+template<typename T> inline void rev(T xs) { std::reverse(all(xs)); }
+template<typename T> inline void sort(T xs) { std::sort(all(xs)); }
+template<typename T> inline void ssort(T xs) { std::stable_sort(all(xs)); }
+template<typename T> inline void unique(T xs) { std::unique(all(xs)); }
+template<typename T> inline void uniq(T xs) { xs.erase( std::unique(all(xs)), xs.end() ); }
+template<typename T, typename U> inline void fill(T xs, U x) { std::fill(all(xs), x); }
+template<typename T, typename U> inline U minim(const T xs) { return *std::min_element(all(xs)); }
+template<typename T, typename U> inline U maxim(const T xs) { return *std::max_element(all(xs)); }
+template<typename T> inline void nextp(T xs) { return std::next_permutation(all(xs)); }
+template<typename T> inline void prevp(T xs) { return std::prev_permutation(all(xs)); }
+template<typename T> inline string tos(T x) { stringstream s; s << x; return s.str(); }
+inline int powi(int a, int b) { return int( std::pow(double(a), double(b)) ); }
+inline int powl(ll a, ll b) { return ll( std::pow(double(a), double(b)) ); }
+template<typename T> inline void pp(const T & x) { cout << x << endl; }
+
+inline ll gcd(ll a, ll b) {
+  if (a < 0 && b < 0) return gcd(-a,-b);
+  if (a == 0) return b;
+  if (b == 0) return a;
+  if (a > b) return gcd(b, a);
+  if (!(b % a)) return a;
+  return gcd(a, b % a);
+}
+
+// TODO merge the following two pieces of code together
+
+template <typename T>
+inline ostream& operator << (ostream& os, const set<T> & xs) {
+  FOR(i,xs) os << i ? ", " : "{ " << xs[i];
+  return os << " }";
+}
+
+template <typename T>
+inline ostream& operator << (ostream& os, const vector<T> & xs) {
+  FOR(i,xs) os << i ? ", " : "{ " << xs[i];
+  return os << " }";
+}
+
+template<class S,class T>
+inline ostream & operator << (ostream & os, const pair<S,T> & a) {
+  os << "(" << a.first << ", " << a.second << ")";
+  return os;
+}
+
+vs split( const string & s, const string & delim = " " ) {
+  vs res;
+  string t;
+  forea(c,s) {
+    if ( delim.find( c ) != string::npos ) {
+      if ( !t.empty() ) {
+        res.pb( t );
+        t = "";
+      }
+    } else {
+      t += c;
+    }
+  }
+  if ( ! t.empty() ) {
+    res.push_back(t);
+  }
+  return res;
+}
+
+vi ints( const str & s, const str & delim = " " ) {
+  vs ss = split( s, delim );
+  vi is;
+  forea(s,ss) is.push_back( atoi( s.c_str() ) );
+  return is;
+}
+
+// following is needed for 'main'
+#define ARRSIZE(x) (sizeof(x)/sizeof(x[0]))
+// $ENDREUSE$
+
+
+
+
+
+
+
+$BEGINCUT$
+template<typename T> void print( T a ) {
+  cerr << a;
+}
+static void print( long long a ) {
+  cerr << a << "L";
+}
+static void print( string a ) {
+  cerr << '"' << a << '"';
+}
+template<typename T> void print( vector<T> a ) {
+  cerr << "{";
+  for ( int i = 0 ; i != a.size() ; i++ ) {
+    if ( i != 0 ) cerr << ", ";
+    print( a[i] );
+  }
+  cerr << "}" << endl;
+}
+template<typename T> void eq( int n, T have, T need ) {
+  if ( have == need ) {
+    cerr << "Case " << n << " passed." << endl;
+  } else {
+    cerr << "Case " << n << " failed: expected ";
+    print( need );
+    cerr << " received ";
+    print( have );
+    cerr << "." << endl;
+  }
+}
+template<typename T> void eq( int n, vector<T> have, vector<T> need ) {
+  if( have.size() != need.size() ) {
+    cerr << "Case " << n << " failed: returned " << have.size() << " elements; expected " << need.size() << " elements." << endl;
+    cerr << "  have: "; print( have );
+    cerr << "  need: "; print( need );
+    return;
+  }
+  for( size_t i= 0; i < have.size(); i++ ) {
+    if( have[i] != need[i] ) {
+      cerr << "Case " << n << " failed. Expected and returned array differ in position " << i << "." << endl;
+      cerr << "  have: "; print( have );
+      cerr << "  need: "; print( need );
+      return;
+    }
+  }
+  cerr << "Case " << n << " passed." << endl;
+}
+static void eq( int n, string have, string need ) {
+  if ( have == need ) {
+    cerr << "Case " << n << " passed." << endl;
+  } else {
+    cerr << "Case " << n << " failed: expected ";
+    print( need );
+    cerr << " received ";
+    print( have );
+    cerr << "." << endl;
+  }
+}
+$ENDCUT$
+
+
+
+
+
+
+
+
+
+
+
+
+
+class $CLASSNAME$ {
+  public:
+  $RC$ $METHODNAME$($METHODPARMS$) {
+    $RC$ res;
+    return res;
+  }
+};
+
+
+
+
+
+
+
+$BEGINCUT$
+int main() {
+$MAINBODY$
+#ifdef WIN32
+  cin.get();
+#endif
+  return 0;
+}
+$ENDCUT$
+


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

[Assorted-commits] SF.net SVN: assorted: [517] numa-bench/trunk/doc/analysis.txt

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

Revision: 517
          http://assorted.svn.sourceforge.net/assorted/?rev=517&view=rev
Author:   yangzhang
Date:     2008-02-26 12:49:46 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
fixed links

Modified Paths:
--------------
    numa-bench/trunk/doc/analysis.txt

Modified: numa-bench/trunk/doc/analysis.txt
===================================================================
--- numa-bench/trunk/doc/analysis.txt	2008-02-26 20:48:46 UTC (rev 516)
+++ numa-bench/trunk/doc/analysis.txt	2008-02-26 20:49:46 UTC (rev 517)
@@ -32,9 +32,9 @@
 - Read vs. write
   - Substantial difference. Random writes are ~2x slower than random reads.
   - Compare
-    [a](graphs/ncores-16-size-1000000000-nreps-1-shuffle-0-par-0-pin-1-local-0-write-0-cross-0.pdf)
+    [a](graphs/ncores-16-size-100000000-nreps-1-shuffle-1-par-0-pin-1-local-0-write-0-cross-0.pdf)
     and
-    [b](graphs/ncores-16-size-1000000000-nreps-1-shuffle-0-par-0-pin-1-local-0-write-1-cross-0.pdf)
+    [b](graphs/ncores-16-size-100000000-nreps-1-shuffle-1-par-0-pin-1-local-0-write-1-cross-0.pdf)
 - Does `malloc` tend to allocate locally?
   - Yes, because working with memory allocated from the current thread shows
     improved times.


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

[Assorted-commits] SF.net SVN: assorted: [516] numa-bench/trunk/src/chew.cc

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

Revision: 516
          http://assorted.svn.sourceforge.net/assorted/?rev=516&view=rev
Author:   yangzhang
Date:     2008-02-26 12:48:46 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
removed unnecessary assertion

Modified Paths:
--------------
    numa-bench/trunk/src/chew.cc

Modified: numa-bench/trunk/src/chew.cc
===================================================================
--- numa-bench/trunk/src/chew.cc	2008-02-26 20:31:16 UTC (rev 515)
+++ numa-bench/trunk/src/chew.cc	2008-02-26 20:48:46 UTC (rev 516)
@@ -231,8 +231,6 @@
        << " nnodes  " << config.nnodes
        << endl;
 
-  checkmsg(RAND_MAX > config.size / sizeof(int), "PRNG range not large enough");
-
   void *p = malloc(config.size);
   check(p != NULL);
 


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

[Assorted-commits] SF.net SVN: assorted: [515] numa-bench/trunk/src/GNUmakefile

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

Revision: 515
          http://assorted.svn.sourceforge.net/assorted/?rev=515&view=rev
Author:   yangzhang
Date:     2008-02-26 12:31:16 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
updated simple-build makefile

Modified Paths:
--------------
    numa-bench/trunk/src/GNUmakefile

Modified: numa-bench/trunk/src/GNUmakefile
===================================================================
--- numa-bench/trunk/src/GNUmakefile	2008-02-26 20:31:00 UTC (rev 514)
+++ numa-bench/trunk/src/GNUmakefile	2008-02-26 20:31:16 UTC (rev 515)
@@ -14,8 +14,9 @@
 FLAGS_0 := -g3
 
 OBJDIR_0 := $(OUTDIR)/$(TARGET_0)-obj
-CFLAGS_0 := $(CFLAGS) $(FLAGS_0)
-CXXFLAGS_0 := $(CXXFLAGS) $(FLAGS_0)
+FLAGS_0 := -Wall -Werror $(FLAGS_0)
+CFLAGS_0 := $(FLAGS_0) $(CFLAGS)
+CXXFLAGS_0 := $(FLAGS_0) $(CXXFLAGS)
 CPPFLAGS_0 := $(CPPFLAGS)
 LDFLAGS_0 := $(LDFLAGS) $(LIBS_0) $(AUTOLIBS_0)
 EXE_0 := $(OUTDIR)/$(TARGET_0)
@@ -71,8 +72,9 @@
 FLAGS_1 := -O3
 
 OBJDIR_1 := $(OUTDIR)/$(TARGET_1)-obj
-CFLAGS_1 := $(CFLAGS) $(FLAGS_1)
-CXXFLAGS_1 := $(CXXFLAGS) $(FLAGS_1)
+FLAGS_1 := -Wall -Werror $(FLAGS_1)
+CFLAGS_1 := $(FLAGS_1) $(CFLAGS)
+CXXFLAGS_1 := $(FLAGS_1) $(CXXFLAGS)
 CPPFLAGS_1 := $(CPPFLAGS)
 LDFLAGS_1 := $(LDFLAGS) $(LIBS_1) $(AUTOLIBS_1)
 EXE_1 := $(OUTDIR)/$(TARGET_1)
@@ -128,8 +130,9 @@
 FLAGS_2 := -pg
 
 OBJDIR_2 := $(OUTDIR)/$(TARGET_2)-obj
-CFLAGS_2 := $(CFLAGS) $(FLAGS_2)
-CXXFLAGS_2 := $(CXXFLAGS) $(FLAGS_2)
+FLAGS_2 := -Wall -Werror $(FLAGS_2)
+CFLAGS_2 := $(FLAGS_2) $(CFLAGS)
+CXXFLAGS_2 := $(FLAGS_2) $(CXXFLAGS)
 CPPFLAGS_2 := $(CPPFLAGS)
 LDFLAGS_2 := $(LDFLAGS) $(LIBS_2) $(AUTOLIBS_2)
 EXE_2 := $(OUTDIR)/$(TARGET_2)
@@ -193,8 +196,9 @@
 FLAGS_3 := -g3
 
 OBJDIR_3 := $(OUTDIR)/$(TARGET_3)-obj
-CFLAGS_3 := $(CFLAGS) $(FLAGS_3)
-CXXFLAGS_3 := $(CXXFLAGS) $(FLAGS_3)
+FLAGS_3 := -Wall -Werror $(FLAGS_3)
+CFLAGS_3 := $(FLAGS_3) $(CFLAGS)
+CXXFLAGS_3 := $(FLAGS_3) $(CXXFLAGS)
 CPPFLAGS_3 := $(CPPFLAGS)
 LDFLAGS_3 := $(LDFLAGS) $(LIBS_3) $(AUTOLIBS_3)
 EXE_3 := $(OUTDIR)/$(TARGET_3)
@@ -250,8 +254,9 @@
 FLAGS_4 := -O3
 
 OBJDIR_4 := $(OUTDIR)/$(TARGET_4)-obj
-CFLAGS_4 := $(CFLAGS) $(FLAGS_4)
-CXXFLAGS_4 := $(CXXFLAGS) $(FLAGS_4)
+FLAGS_4 := -Wall -Werror $(FLAGS_4)
+CFLAGS_4 := $(FLAGS_4) $(CFLAGS)
+CXXFLAGS_4 := $(FLAGS_4) $(CXXFLAGS)
 CPPFLAGS_4 := $(CPPFLAGS)
 LDFLAGS_4 := $(LDFLAGS) $(LIBS_4) $(AUTOLIBS_4)
 EXE_4 := $(OUTDIR)/$(TARGET_4)
@@ -307,8 +312,9 @@
 FLAGS_5 := -pg
 
 OBJDIR_5 := $(OUTDIR)/$(TARGET_5)-obj
-CFLAGS_5 := $(CFLAGS) $(FLAGS_5)
-CXXFLAGS_5 := $(CXXFLAGS) $(FLAGS_5)
+FLAGS_5 := -Wall -Werror $(FLAGS_5)
+CFLAGS_5 := $(FLAGS_5) $(CFLAGS)
+CXXFLAGS_5 := $(FLAGS_5) $(CXXFLAGS)
 CPPFLAGS_5 := $(CPPFLAGS)
 LDFLAGS_5 := $(LDFLAGS) $(LIBS_5) $(AUTOLIBS_5)
 EXE_5 := $(OUTDIR)/$(TARGET_5)


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

[Assorted-commits] SF.net SVN: assorted: [514] numa-bench/trunk/src

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

Revision: 514
          http://assorted.svn.sourceforge.net/assorted/?rev=514&view=rev
Author:   yangzhang
Date:     2008-02-26 12:31:00 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
added round-robin node allocation; renamed cores to cpus

Modified Paths:
--------------
    numa-bench/trunk/src/chew.bash
    numa-bench/trunk/src/chew.cc

Modified: numa-bench/trunk/src/chew.bash
===================================================================
--- numa-bench/trunk/src/chew.bash	2008-02-26 19:48:37 UTC (rev 513)
+++ numa-bench/trunk/src/chew.bash	2008-02-26 20:31:00 UTC (rev 514)
@@ -12,32 +12,32 @@
 
 KB=000 MB=000000 GB=000000000
 
-#   ncores    size nreps shuffle par pin local write cross
+#    ncpus    size nreps shuffle par pin local write cross rrnodes
 
 echo writes
-run     16  100$MB     1       0   0   1     0     1     0
-run     16 1000$MB     1       0   0   1     0     1     0
-run     16  100$MB    10       0   0   1     0     1     0
-run     16  100$MB     1       1   0   1     0     1     0
+run     16  100$MB     1       0   0   1     0     1     0       1
+run     16 1000$MB     1       0   0   1     0     1     0       1
+run     16  100$MB    10       0   0   1     0     1     0       1
+run     16  100$MB     1       1   0   1     0     1     0       1
 
 echo reads
-run     16 1000$MB     1       0   0   1     0     0     0
-run     16  100$MB     1       1   0   1     0     0     0
+run     16 1000$MB     1       0   0   1     0     0     0       1
+run     16  100$MB     1       1   0   1     0     0     0       1
 
 for n in 1 2 4 8 12 16 ; do
   echo par
-  run   $n   10$MB     1       0   1   1     0     0     0
-  run   $n   10$MB     1       1   1   1     0     0     0
-  run   $n   10$MB     1       0   1   1     1     0     0
-  run   $n   10$MB     1       1   1   1     1     0     0
-  run   $n   10$MB     1       0   1   1     0     1     0
-  run   $n   10$MB     1       1   1   1     0     1     0
-  run   $n   10$MB     1       0   1   1     1     1     0
-  run   $n   10$MB     1       1   1   1     1     1     0
+  run   $n   10$MB     1       0   1   1     0     0     0       1
+  run   $n   10$MB     1       1   1   1     0     0     0       1
+  run   $n   10$MB     1       0   1   1     1     0     0       1
+  run   $n   10$MB     1       1   1   1     1     0     0       1
+  run   $n   10$MB     1       0   1   1     0     1     0       1
+  run   $n   10$MB     1       1   1   1     0     1     0       1
+  run   $n   10$MB     1       0   1   1     1     1     0       1
+  run   $n   10$MB     1       1   1   1     1     1     0       1
 
   echo cross
-  run   $n   10$MB     1       0   1   1     0     0     1
-  run   $n   10$MB     1       1   1   1     0     0     1
-  run   $n   10$MB     1       0   1   1     0     1     1
-  run   $n   10$MB     1       1   1   1     0     1     1
+  run   $n   10$MB     1       0   1   1     0     0     1       1
+  run   $n   10$MB     1       1   1   1     0     0     1       1
+  run   $n   10$MB     1       0   1   1     0     1     1       1
+  run   $n   10$MB     1       1   1   1     0     1     1       1
 done

Modified: numa-bench/trunk/src/chew.cc
===================================================================
--- numa-bench/trunk/src/chew.cc	2008-02-26 19:48:37 UTC (rev 513)
+++ numa-bench/trunk/src/chew.cc	2008-02-26 20:31:00 UTC (rev 514)
@@ -20,12 +20,12 @@
 struct config
 {
   /**
-   * The number of cores to test. This is a parameter (rather than
-   * auto-detected) because it additionally serves to mean the number of cores
+   * The number of CPUs to test. This is a parameter (rather than
+   * auto-detected) because it additionally serves to mean the number of CPUs
    * we want to test in parallel. As this program evolves, these may be
    * separated.
    */
-  const int ncores;
+  const unsigned int ncpus;
 
   /**
    * Size in bytes of the buffer to chew.
@@ -35,7 +35,7 @@
   /**
    * Number of repetitions to chew.
    */
-  const int nreps;
+  const unsigned int nreps;
 
   /**
    * Perform rand access, otherwise sequential scan.
@@ -43,12 +43,13 @@
   const bool shuffle;
 
   /**
-   * Chew in parallel, otherwise each core chews serially.
+   * Chew in parallel, otherwise each worker chews serially.
    */
   const bool par;
 
   /**
-   * Pin thread i to core i, otherwise let the OS manage things.
+   * Pin workers to CPUs (which CPU determined by whether we're using
+   * round-robin CPU allocation), otherwise let the OS manage things.
    */
   const bool pin;
 
@@ -67,6 +68,17 @@
    * global/local buffer.
    */
   const bool cross;
+
+  /**
+   * When using fewer than the maximum number of CPUs, allocate nodes in
+   * round-robin fashion.
+   */
+  const bool rrnodes;
+
+  /**
+   * The number of nodes on this machine.
+   */
+  const unsigned int nnodes;
 };
 
 void*** partitions;
@@ -129,15 +141,21 @@
 
 /**
  * \param pp The start of the buffer to chew.
- * \param cpu Which CPU to pin our thread to.
+ * \param worker From this we can determine which CPU to pin our thread to.
  * \param config The experiment configuration parameters.
  * \param label Prefix for the elapsed time output.
  */
 void*
-chew(void* pp, unsigned int cpu, const config & config, bool warmup)
+chew(void* pp, unsigned int worker, const config & config, bool warmup)
 {
-  // Pin this thread to cpu `cpu`.
+  // Pin this thread to a CPU.
   if (config.pin) {
+    unsigned int ncpus = config.ncpus,
+                 nnodes = config.nnodes,
+                 cpuspernode = ncpus / nnodes,
+                 cpu = !config.rrnodes ?
+                       worker :
+                       (cpuspernode * worker) % ncpus + worker / cpuspernode;
     pin_thread(cpu);
   }
 
@@ -145,16 +163,17 @@
   timer t(": ");
 
   if (!warmup && config.cross) {
-    size_t len = config.size / config.ncores;
-    for (int i = 0; i < config.ncores; i++) {
-      partitions[cpu][i] = new char[len];
+    size_t len = config.size / config.ncpus;
+    for (unsigned int i = 0; i < config.ncpus; i++) {
+      partitions[worker][i] = new char[len];
     }
     int barrier_result = pthread_barrier_wait(&cross_barrier);
-    check(barrier_result == PTHREAD_BARRIER_SERIAL_THREAD || barrier_result == 0);
+    check(barrier_result == PTHREAD_BARRIER_SERIAL_THREAD ||
+          barrier_result == 0);
     // TODO: make this more interesting than just a sequential traversal over
     // the partitions.
-    for (int i = 0; i < config.ncores; i++) {
-      chew1(partitions[i][cpu], config, len);
+    for (unsigned int i = 0; i < config.ncpus; i++) {
+      chew1(partitions[i][worker], config, len);
     }
   } else {
     chew1(p, config, config.size);
@@ -162,7 +181,7 @@
 
   // Print the elapsed time and "result".
   if (warmup) cout << "warmup: ";
-  cout << cpu;
+  cout << worker;
   t.print();
 
   if (config.local) free(p);
@@ -176,27 +195,30 @@
   // So that our global shared malloc takes place on the CPU 0's node.
   pin_thread(0);
 
-  if (argc < 10) {
+  if (argc < 11) {
     cerr << argv[0] <<
-      " <ncores> <size> <nreps> <shuffle> <par> <pin> <local> <write>" << endl;
+      " <ncpus> <size> <nreps> <shuffle> <par> <pin>"
+      " <local> <write> <rrnodes> <nnodes>" << endl;
     return 1;
   }
 
-  // Parse command-line arguments. TODO
-  const config config = { 
-    atoi(argv[1]),
-    atoi(argv[2]),
-    atoi(argv[3]),
-    atoi(argv[4]),
-    atoi(argv[5]),
-    atoi(argv[6]),
-    atoi(argv[7]),
-    atoi(argv[8]),
-    atoi(argv[9])
+  // Parse command-line arguments.
+  const config config = {
+    atoi(argv[ 1]),
+    atoi(argv[ 2]),
+    atoi(argv[ 3]),
+    atoi(argv[ 4]),
+    atoi(argv[ 5]),
+    atoi(argv[ 6]),
+    atoi(argv[ 7]),
+    atoi(argv[ 8]),
+    atoi(argv[ 9]),
+    atoi(argv[10]),
+    atoi(argv[11])
   };
 
   cout << "config:"
-       << " ncores "  << config.ncores
+       << " ncpus "   << config.ncpus
        << " size "    << config.size
        << " nreps "   << config.nreps
        << " shuffle " << config.shuffle
@@ -204,7 +226,10 @@
        << " pin "     << config.pin
        << " local "   << config.local
        << " write "   << config.write
-       << " cross "   << config.cross << endl;
+       << " cross "   << config.cross
+       << " rrnodes " << config.rrnodes
+       << " nnodes  " << config.nnodes
+       << endl;
 
   checkmsg(RAND_MAX > config.size / sizeof(int), "PRNG range not large enough");
 
@@ -212,9 +237,9 @@
   check(p != NULL);
 
   if (config.cross) {
-    partitions = new void**[config.ncores];
-    for (unsigned int i  = 0; i < config.ncores; i++)
-      partitions[i] = new void*[config.ncores];
+    partitions = new void**[config.ncpus];
+    for (unsigned int i  = 0; i < config.ncpus; i++)
+      partitions[i] = new void*[config.ncpus];
   }
 
   // Warmup.
@@ -222,18 +247,18 @@
 
   if (config.par) {
     // Chew the memory area from each core in parallel (and also chew own).
-    pthread_t ts[config.ncores];
-    check(0 == pthread_barrier_init(&cross_barrier, NULL, config.ncores));
-    for (int i = 0; i < config.ncores; i++) {
+    pthread_t ts[config.ncpus];
+    check(0 == pthread_barrier_init(&cross_barrier, NULL, config.ncpus));
+    for (unsigned int i = 0; i < config.ncpus; i++) {
       ts[i] = spawn(bind(chew, p, i, ref(config), false));
     }
-    for (int i = 0; i < config.ncores; i++) {
+    for (unsigned int i = 0; i < config.ncpus; i++) {
       check(pthread_join(ts[i], NULL) == 0);
     }
     check(0 == pthread_barrier_destroy(&cross_barrier));
   } else {
     // Chew the memory area from each core in sequence.
-    for (int i = 0; i < config.ncores; i++) {
+    for (unsigned int i = 0; i < config.ncpus; i++) {
       chew(p, i, config, false);
     }
   }


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

[Assorted-commits] SF.net SVN: assorted: [513] numa-bench/trunk/src

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

Revision: 513
          http://assorted.svn.sourceforge.net/assorted/?rev=513&view=rev
Author:   yangzhang
Date:     2008-02-26 11:48:37 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
renamed malloc to chew

Modified Paths:
--------------
    numa-bench/trunk/src/CMakeLists.txt
    numa-bench/trunk/src/GNUmakefile
    numa-bench/trunk/src/Jamroot
    numa-bench/trunk/src/build

Added Paths:
-----------
    numa-bench/trunk/src/chew.bash
    numa-bench/trunk/src/chew.cc

Removed Paths:
-------------
    numa-bench/trunk/src/malloc.bash
    numa-bench/trunk/src/malloc.cc

Modified: numa-bench/trunk/src/CMakeLists.txt
===================================================================
--- numa-bench/trunk/src/CMakeLists.txt	2008-02-26 19:41:17 UTC (rev 512)
+++ numa-bench/trunk/src/CMakeLists.txt	2008-02-26 19:48:37 UTC (rev 513)
@@ -1,7 +1,7 @@
 project (numa-bench)
 
-add_executable (malloc malloc.cc)
-target_link_libraries (malloc pthread)
+add_executable (chew chew.cc)
+target_link_libraries (chew pthread)
 
 add_executable (avail avail.cc)
 target_link_libraries (avail numa)

Modified: numa-bench/trunk/src/GNUmakefile
===================================================================
--- numa-bench/trunk/src/GNUmakefile	2008-02-26 19:41:17 UTC (rev 512)
+++ numa-bench/trunk/src/GNUmakefile	2008-02-26 19:48:37 UTC (rev 513)
@@ -1,16 +1,16 @@
 OUTDIR := /home/yang/work/assorted/numa-bench/trunk/src/out
 SRCDIR := /home/yang/work/assorted/numa-bench/trunk/src
 SRCPATH := /home/yang/work/assorted/numa-bench/trunk/src
-all: malloc avail
-PREFIX_0 := /opt/malloc
+all: chew avail
+PREFIX_0 := /opt/chew
 BINDIR_0 := $(PREFIX_0)/bin
 
 LIBS_0 := -lpthread
-SRCS_0 := $(SRCDIR)/malloc.cc
-ORIGTARGET_0 := malloc
+SRCS_0 := $(SRCDIR)/chew.cc
+ORIGTARGET_0 := chew
 AUTOLIBS_0 :=  
 LANG_0 := cpp
-TARGET_0 := malloc-dbg
+TARGET_0 := chew-dbg
 FLAGS_0 := -g3
 
 OBJDIR_0 := $(OUTDIR)/$(TARGET_0)-obj
@@ -59,15 +59,15 @@
 $(ORIGTARGET_0): $(TARGET_0)
 
 .PHONY: $(ORIGTARGET_0)
-PREFIX_1 := /opt/malloc
+PREFIX_1 := /opt/chew
 BINDIR_1 := $(PREFIX_1)/bin
 
 LIBS_1 := -lpthread
-SRCS_1 := $(SRCDIR)/malloc.cc
-ORIGTARGET_1 := malloc
+SRCS_1 := $(SRCDIR)/chew.cc
+ORIGTARGET_1 := chew
 AUTOLIBS_1 :=  
 LANG_1 := cpp
-TARGET_1 := malloc-opt
+TARGET_1 := chew-opt
 FLAGS_1 := -O3
 
 OBJDIR_1 := $(OUTDIR)/$(TARGET_1)-obj
@@ -116,15 +116,15 @@
 $(ORIGTARGET_1): $(TARGET_1)
 
 .PHONY: $(ORIGTARGET_1)
-PREFIX_2 := /opt/malloc
+PREFIX_2 := /opt/chew
 BINDIR_2 := $(PREFIX_2)/bin
 
 LIBS_2 := -lpthread
-SRCS_2 := $(SRCDIR)/malloc.cc
-ORIGTARGET_2 := malloc
+SRCS_2 := $(SRCDIR)/chew.cc
+ORIGTARGET_2 := chew
 AUTOLIBS_2 :=  
 LANG_2 := cpp
-TARGET_2 := malloc-gprof
+TARGET_2 := chew-gprof
 FLAGS_2 := -pg
 
 OBJDIR_2 := $(OUTDIR)/$(TARGET_2)-obj

Modified: numa-bench/trunk/src/Jamroot
===================================================================
--- numa-bench/trunk/src/Jamroot	2008-02-26 19:41:17 UTC (rev 512)
+++ numa-bench/trunk/src/Jamroot	2008-02-26 19:48:37 UTC (rev 513)
@@ -1,2 +1,2 @@
 lib pthread ;
-exe malloc : malloc.cc pthread : <toolset>gcc:<cxxflags>-g3 ;
+exe chew : chew.cc pthread : <toolset>gcc:<cxxflags>-g3 ;

Modified: numa-bench/trunk/src/build
===================================================================
--- numa-bench/trunk/src/build	2008-02-26 19:41:17 UTC (rev 512)
+++ numa-bench/trunk/src/build	2008-02-26 19:48:37 UTC (rev 513)
@@ -1,5 +1,5 @@
-malloc:
-  srcs: [malloc.cc]
+chew:
+  srcs: [chew.cc]
   libs: [pthread]
 
 avail:

Copied: numa-bench/trunk/src/chew.bash (from rev 511, numa-bench/trunk/src/malloc.bash)
===================================================================
--- numa-bench/trunk/src/chew.bash	                        (rev 0)
+++ numa-bench/trunk/src/chew.bash	2008-02-26 19:48:37 UTC (rev 513)
@@ -0,0 +1,43 @@
+#!/usr/bin/env bash
+
+set -o errexit -o nounset
+
+make -s chew-opt
+
+function run {
+  for i in {1..3}
+  do out/chew-opt "$@"
+  done
+}
+
+KB=000 MB=000000 GB=000000000
+
+#   ncores    size nreps shuffle par pin local write cross
+
+echo writes
+run     16  100$MB     1       0   0   1     0     1     0
+run     16 1000$MB     1       0   0   1     0     1     0
+run     16  100$MB    10       0   0   1     0     1     0
+run     16  100$MB     1       1   0   1     0     1     0
+
+echo reads
+run     16 1000$MB     1       0   0   1     0     0     0
+run     16  100$MB     1       1   0   1     0     0     0
+
+for n in 1 2 4 8 12 16 ; do
+  echo par
+  run   $n   10$MB     1       0   1   1     0     0     0
+  run   $n   10$MB     1       1   1   1     0     0     0
+  run   $n   10$MB     1       0   1   1     1     0     0
+  run   $n   10$MB     1       1   1   1     1     0     0
+  run   $n   10$MB     1       0   1   1     0     1     0
+  run   $n   10$MB     1       1   1   1     0     1     0
+  run   $n   10$MB     1       0   1   1     1     1     0
+  run   $n   10$MB     1       1   1   1     1     1     0
+
+  echo cross
+  run   $n   10$MB     1       0   1   1     0     0     1
+  run   $n   10$MB     1       1   1   1     0     0     1
+  run   $n   10$MB     1       0   1   1     0     1     1
+  run   $n   10$MB     1       1   1   1     0     1     1
+done

Copied: numa-bench/trunk/src/chew.cc (from rev 512, numa-bench/trunk/src/malloc.cc)
===================================================================
--- numa-bench/trunk/src/chew.cc	                        (rev 0)
+++ numa-bench/trunk/src/chew.cc	2008-02-26 19:48:37 UTC (rev 513)
@@ -0,0 +1,246 @@
+#include <fstream>
+#include <iostream>
+
+#include <sched.h>
+
+#include <boost/bind.hpp>
+
+#include <commons/check.h>
+#include <commons/rand.h>
+#include <commons/threads.h>
+#include <commons/time.h>
+#include <commons/boost/threads.h>
+
+using namespace boost;
+using namespace commons;
+using namespace std;
+
+pthread_barrier_t cross_barrier;
+
+struct config
+{
+  /**
+   * The number of cores to test. This is a parameter (rather than
+   * auto-detected) because it additionally serves to mean the number of cores
+   * we want to test in parallel. As this program evolves, these may be
+   * separated.
+   */
+  const int ncores;
+
+  /**
+   * Size in bytes of the buffer to chew.
+   */
+  const size_t size;
+
+  /**
+   * Number of repetitions to chew.
+   */
+  const int nreps;
+
+  /**
+   * Perform rand access, otherwise sequential scan.
+   */
+  const bool shuffle;
+
+  /**
+   * Chew in parallel, otherwise each core chews serially.
+   */
+  const bool par;
+
+  /**
+   * Pin thread i to core i, otherwise let the OS manage things.
+   */
+  const bool pin;
+
+  /**
+   * Chew my own memory, otherwise chew the given (shared) memory.
+   */
+  const bool local;
+
+  /**
+   * Do writes, otherwise just do reads.
+   */
+  const bool write;
+
+  /**
+   * Test cross-communication (use partitions), otherwise use either the
+   * global/local buffer.
+   */
+  const bool cross;
+};
+
+void*** partitions;
+int global_sum;
+
+/**
+ * \param p The buffer to chew.
+ * \param config The experiment configuration.
+ * \param len Length of the buffer.
+ */
+void
+chew1(void* pp, config config, size_t len)
+{
+  int* p = (int*) pp;
+  const size_t count = len / sizeof(int);
+  int sum = 0;
+  posix_rand rand(current_time_millis() ^ gettid());
+  if (config.write) {
+    // Write to the region.
+    if (config.shuffle) {
+      // Random access into the memory region.
+      for (unsigned int c = 0; c < config.nreps; c++) {
+        for (size_t i = 0; i < count; i++) {
+          // NOTE: Using r as the index assumes that rand generates large-enough
+          // values.
+          int r = rand();
+          sum += p[r % count] += r;
+        }
+      }
+    } else {
+      // Sequential scan through the memory region.
+      for (unsigned int c = 0; c < config.nreps; c++) {
+        for (size_t i = 0; i < count; i++) {
+          sum += p[i] += rand();
+        }
+      }
+    }
+  } else {
+    // Only read from the region.
+    if (config.shuffle) {
+      // Random access into the memory region.
+      for (unsigned int c = 0; c < config.nreps; c++) {
+        for (size_t i = 0; i < count; i++) {
+          // NOTE: Using r as the index assumes that rand generates large-enough
+          // values.
+          sum += p[rand() % count];
+        }
+      }
+    } else {
+      // Sequential scan through the memory region.
+      for (unsigned int c = 0; c < config.nreps; c++) {
+        for (size_t i = 0; i < count; i++) {
+          sum += p[i] + rand();
+        }
+      }
+    }
+  }
+  global_sum += sum;
+}
+
+/**
+ * \param pp The start of the buffer to chew.
+ * \param cpu Which CPU to pin our thread to.
+ * \param config The experiment configuration parameters.
+ * \param label Prefix for the elapsed time output.
+ */
+void*
+chew(void* pp, unsigned int cpu, const config & config, bool warmup)
+{
+  // Pin this thread to cpu `cpu`.
+  if (config.pin) {
+    pin_thread(cpu);
+  }
+
+  void* p = config.local ? malloc(config.size) : pp;
+  timer t(": ");
+
+  if (!warmup && config.cross) {
+    size_t len = config.size / config.ncores;
+    for (int i = 0; i < config.ncores; i++) {
+      partitions[cpu][i] = new char[len];
+    }
+    int barrier_result = pthread_barrier_wait(&cross_barrier);
+    check(barrier_result == PTHREAD_BARRIER_SERIAL_THREAD || barrier_result == 0);
+    // TODO: make this more interesting than just a sequential traversal over
+    // the partitions.
+    for (int i = 0; i < config.ncores; i++) {
+      chew1(partitions[i][cpu], config, len);
+    }
+  } else {
+    chew1(p, config, config.size);
+  }
+
+  // Print the elapsed time and "result".
+  if (warmup) cout << "warmup: ";
+  cout << cpu;
+  t.print();
+
+  if (config.local) free(p);
+
+  return NULL;
+}
+
+int
+main(int argc, char** argv)
+{
+  // So that our global shared malloc takes place on the CPU 0's node.
+  pin_thread(0);
+
+  if (argc < 10) {
+    cerr << argv[0] <<
+      " <ncores> <size> <nreps> <shuffle> <par> <pin> <local> <write>" << endl;
+    return 1;
+  }
+
+  // Parse command-line arguments. TODO
+  const config config = { 
+    atoi(argv[1]),
+    atoi(argv[2]),
+    atoi(argv[3]),
+    atoi(argv[4]),
+    atoi(argv[5]),
+    atoi(argv[6]),
+    atoi(argv[7]),
+    atoi(argv[8]),
+    atoi(argv[9])
+  };
+
+  cout << "config:"
+       << " ncores "  << config.ncores
+       << " size "    << config.size
+       << " nreps "   << config.nreps
+       << " shuffle " << config.shuffle
+       << " par "     << config.par
+       << " pin "     << config.pin
+       << " local "   << config.local
+       << " write "   << config.write
+       << " cross "   << config.cross << endl;
+
+  checkmsg(RAND_MAX > config.size / sizeof(int), "PRNG range not large enough");
+
+  void *p = malloc(config.size);
+  check(p != NULL);
+
+  if (config.cross) {
+    partitions = new void**[config.ncores];
+    for (unsigned int i  = 0; i < config.ncores; i++)
+      partitions[i] = new void*[config.ncores];
+  }
+
+  // Warmup.
+  chew(p, 0, config, true);
+
+  if (config.par) {
+    // Chew the memory area from each core in parallel (and also chew own).
+    pthread_t ts[config.ncores];
+    check(0 == pthread_barrier_init(&cross_barrier, NULL, config.ncores));
+    for (int i = 0; i < config.ncores; i++) {
+      ts[i] = spawn(bind(chew, p, i, ref(config), false));
+    }
+    for (int i = 0; i < config.ncores; i++) {
+      check(pthread_join(ts[i], NULL) == 0);
+    }
+    check(0 == pthread_barrier_destroy(&cross_barrier));
+  } else {
+    // Chew the memory area from each core in sequence.
+    for (int i = 0; i < config.ncores; i++) {
+      chew(p, i, config, false);
+    }
+  }
+
+  free(p);
+  ofstream trash("/dev/null");
+  trash << "result: " << global_sum << endl;
+
+  return 0;
+}

Deleted: numa-bench/trunk/src/malloc.bash
===================================================================
--- numa-bench/trunk/src/malloc.bash	2008-02-26 19:41:17 UTC (rev 512)
+++ numa-bench/trunk/src/malloc.bash	2008-02-26 19:48:37 UTC (rev 513)
@@ -1,43 +0,0 @@
-#!/usr/bin/env bash
-
-set -o errexit -o nounset
-
-make -s malloc-opt
-
-function run {
-  for i in {1..3}
-  do out/malloc-opt "$@"
-  done
-}
-
-KB=000 MB=000000 GB=000000000
-
-#   ncores    size nreps shuffle par pin local write cross
-
-echo writes
-run     16  100$MB     1       0   0   1     0     1     0
-run     16 1000$MB     1       0   0   1     0     1     0
-run     16  100$MB    10       0   0   1     0     1     0
-run     16  100$MB     1       1   0   1     0     1     0
-
-echo reads
-run     16 1000$MB     1       0   0   1     0     0     0
-run     16  100$MB     1       1   0   1     0     0     0
-
-for n in 1 2 4 8 12 16 ; do
-  echo par
-  run   $n   10$MB     1       0   1   1     0     0     0
-  run   $n   10$MB     1       1   1   1     0     0     0
-  run   $n   10$MB     1       0   1   1     1     0     0
-  run   $n   10$MB     1       1   1   1     1     0     0
-  run   $n   10$MB     1       0   1   1     0     1     0
-  run   $n   10$MB     1       1   1   1     0     1     0
-  run   $n   10$MB     1       0   1   1     1     1     0
-  run   $n   10$MB     1       1   1   1     1     1     0
-
-  echo cross
-  run   $n   10$MB     1       0   1   1     0     0     1
-  run   $n   10$MB     1       1   1   1     0     0     1
-  run   $n   10$MB     1       0   1   1     0     1     1
-  run   $n   10$MB     1       1   1   1     0     1     1
-done

Deleted: numa-bench/trunk/src/malloc.cc
===================================================================
--- numa-bench/trunk/src/malloc.cc	2008-02-26 19:41:17 UTC (rev 512)
+++ numa-bench/trunk/src/malloc.cc	2008-02-26 19:48:37 UTC (rev 513)
@@ -1,246 +0,0 @@
-#include <fstream>
-#include <iostream>
-
-#include <sched.h>
-
-#include <boost/bind.hpp>
-
-#include <commons/check.h>
-#include <commons/rand.h>
-#include <commons/threads.h>
-#include <commons/time.h>
-#include <commons/boost/threads.h>
-
-using namespace boost;
-using namespace commons;
-using namespace std;
-
-pthread_barrier_t cross_barrier;
-
-struct config
-{
-  /**
-   * The number of cores to test. This is a parameter (rather than
-   * auto-detected) because it additionally serves to mean the number of cores
-   * we want to test in parallel. As this program evolves, these may be
-   * separated.
-   */
-  const int ncores;
-
-  /**
-   * Size in bytes of the buffer to chew.
-   */
-  const size_t size;
-
-  /**
-   * Number of repetitions to chew.
-   */
-  const int nreps;
-
-  /**
-   * Perform rand access, otherwise sequential scan.
-   */
-  const bool shuffle;
-
-  /**
-   * Chew in parallel, otherwise each core chews serially.
-   */
-  const bool par;
-
-  /**
-   * Pin thread i to core i, otherwise let the OS manage things.
-   */
-  const bool pin;
-
-  /**
-   * Chew my own memory, otherwise chew the given (shared) memory.
-   */
-  const bool local;
-
-  /**
-   * Do writes, otherwise just do reads.
-   */
-  const bool write;
-
-  /**
-   * Test cross-communication (use partitions), otherwise use either the
-   * global/local buffer.
-   */
-  const bool cross;
-};
-
-void*** partitions;
-int global_sum;
-
-/**
- * \param p The buffer to chew.
- * \param config The experiment configuration.
- * \param len Length of the buffer.
- */
-void
-chew1(void* pp, config config, size_t len)
-{
-  int* p = (int*) pp;
-  const size_t count = len / sizeof(int);
-  int sum = 0;
-  posix_rand rand(current_time_millis() ^ gettid());
-  if (config.write) {
-    // Write to the region.
-    if (config.shuffle) {
-      // Random access into the memory region.
-      for (unsigned int c = 0; c < config.nreps; c++) {
-        for (size_t i = 0; i < count; i++) {
-          // NOTE: Using r as the index assumes that rand generates large-enough
-          // values.
-          int r = rand();
-          sum += p[r % count] += r;
-        }
-      }
-    } else {
-      // Sequential scan through the memory region.
-      for (unsigned int c = 0; c < config.nreps; c++) {
-        for (size_t i = 0; i < count; i++) {
-          sum += p[i] += rand();
-        }
-      }
-    }
-  } else {
-    // Only read from the region.
-    if (config.shuffle) {
-      // Random access into the memory region.
-      for (unsigned int c = 0; c < config.nreps; c++) {
-        for (size_t i = 0; i < count; i++) {
-          // NOTE: Using r as the index assumes that rand generates large-enough
-          // values.
-          sum += p[rand() % count];
-        }
-      }
-    } else {
-      // Sequential scan through the memory region.
-      for (unsigned int c = 0; c < config.nreps; c++) {
-        for (size_t i = 0; i < count; i++) {
-          sum += p[i] + rand();
-        }
-      }
-    }
-  }
-  global_sum += sum;
-}
-
-/**
- * \param pp The start of the buffer to chew.
- * \param cpu Which CPU to pin our thread to.
- * \param config The experiment configuration parameters.
- * \param label Prefix for the elapsed time output.
- */
-void*
-chew(void* pp, unsigned int cpu, const config & config, bool warmup)
-{
-  // Pin this thread to cpu `cpu`.
-  if (config.pin) {
-    pin_thread(cpu);
-  }
-
-  void* p = config.local ? malloc(config.size) : pp;
-  timer t(": ");
-
-  if (!warmup && config.cross) {
-    size_t len = config.size / config.ncores;
-    for (int i = 0; i < config.ncores; i++) {
-      partitions[cpu][i] = new char[len];
-    }
-    int barrier_result = pthread_barrier_wait(&cross_barrier);
-    check(barrier_result == PTHREAD_BARRIER_SERIAL_THREAD || barrier_result == 0);
-    // TODO: make this more interesting than just a sequential traversal over
-    // the partitions.
-    for (int i = 0; i < config.ncores; i++) {
-      chew1(partitions[i][cpu], config, len);
-    }
-  } else {
-    chew1(p, config, config.size);
-  }
-
-  // Print the elapsed time and "result".
-  if (warmup) cout << "warmup: ";
-  cout << cpu;
-  t.print();
-
-  if (config.local) free(p);
-
-  return NULL;
-}
-
-int
-main(int argc, char** argv)
-{
-  // So that our global shared malloc takes place on the CPU 0's node.
-  pin_thread(0);
-
-  if (argc < 10) {
-    cerr << argv[0] <<
-      " <ncores> <size> <nreps> <shuffle> <par> <pin> <local> <write>" << endl;
-    return 1;
-  }
-
-  // Parse command-line arguments. TODO
-  const config config = { 
-    atoi(argv[1]),
-    atoi(argv[2]),
-    atoi(argv[3]),
-    atoi(argv[4]),
-    atoi(argv[5]),
-    atoi(argv[6]),
-    atoi(argv[7]),
-    atoi(argv[8]),
-    atoi(argv[9])
-  };
-
-  cout << "config:"
-       << " ncores "  << config.ncores
-       << " size "    << config.size
-       << " nreps "   << config.nreps
-       << " shuffle " << config.shuffle
-       << " par "     << config.par
-       << " pin "     << config.pin
-       << " local "   << config.local
-       << " write "   << config.write
-       << " cross "   << config.cross << endl;
-
-  checkmsg(RAND_MAX > config.size / sizeof(int), "PRNG range not large enough");
-
-  void *p = malloc(config.size);
-  check(p != NULL);
-
-  if (config.cross) {
-    partitions = new void**[config.ncores];
-    for (unsigned int i  = 0; i < config.ncores; i++)
-      partitions[i] = new void*[config.ncores];
-  }
-
-  // Warmup.
-  chew(p, 0, config, true);
-
-  if (config.par) {
-    // Chew the memory area from each core in parallel (and also chew own).
-    pthread_t ts[config.ncores];
-    check(0 == pthread_barrier_init(&cross_barrier, NULL, config.ncores));
-    for (int i = 0; i < config.ncores; i++) {
-      ts[i] = spawn(bind(chew, p, i, ref(config), false));
-    }
-    for (int i = 0; i < config.ncores; i++) {
-      check(pthread_join(ts[i], NULL) == 0);
-    }
-    check(0 == pthread_barrier_destroy(&cross_barrier));
-  } else {
-    // Chew the memory area from each core in sequence.
-    for (int i = 0; i < config.ncores; i++) {
-      chew(p, i, config, false);
-    }
-  }
-
-  free(p);
-  ofstream trash("/dev/null");
-  trash << "result: " << global_sum << endl;
-
-  return 0;
-}


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

[Assorted-commits] SF.net SVN: assorted: [512] numa-bench/trunk/src/malloc.cc

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

Revision: 512
          http://assorted.svn.sourceforge.net/assorted/?rev=512&view=rev
Author:   yangzhang
Date:     2008-02-26 11:41:17 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
added custom rng

Modified Paths:
--------------
    numa-bench/trunk/src/malloc.cc

Modified: numa-bench/trunk/src/malloc.cc
===================================================================
--- numa-bench/trunk/src/malloc.cc	2008-02-26 19:41:06 UTC (rev 511)
+++ numa-bench/trunk/src/malloc.cc	2008-02-26 19:41:17 UTC (rev 512)
@@ -1,4 +1,3 @@
-#include <cstdlib>
 #include <fstream>
 #include <iostream>
 
@@ -7,6 +6,7 @@
 #include <boost/bind.hpp>
 
 #include <commons/check.h>
+#include <commons/rand.h>
 #include <commons/threads.h>
 #include <commons/time.h>
 #include <commons/boost/threads.h>
@@ -83,6 +83,7 @@
   int* p = (int*) pp;
   const size_t count = len / sizeof(int);
   int sum = 0;
+  posix_rand rand(current_time_millis() ^ gettid());
   if (config.write) {
     // Write to the region.
     if (config.shuffle) {


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

[Assorted-commits] SF.net SVN: assorted: [511] numa-bench/trunk/src/malloc.bash

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

Revision: 511
          http://assorted.svn.sourceforge.net/assorted/?rev=511&view=rev
Author:   yangzhang
Date:     2008-02-26 11:41:06 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
updated script to use simple-build output path

Modified Paths:
--------------
    numa-bench/trunk/src/malloc.bash

Modified: numa-bench/trunk/src/malloc.bash
===================================================================
--- numa-bench/trunk/src/malloc.bash	2008-02-26 19:40:47 UTC (rev 510)
+++ numa-bench/trunk/src/malloc.bash	2008-02-26 19:41:06 UTC (rev 511)
@@ -2,11 +2,11 @@
 
 set -o errexit -o nounset
 
-make -s malloc
+make -s malloc-opt
 
 function run {
   for i in {1..3}
-  do ./malloc "$@"
+  do out/malloc-opt "$@"
   done
 }
 


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

[Assorted-commits] SF.net SVN: assorted: [510] numa-bench/trunk/src/GNUmakefile

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

Revision: 510
          http://assorted.svn.sourceforge.net/assorted/?rev=510&view=rev
Author:   yangzhang
Date:     2008-02-26 11:40:47 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
added output of simple-build (for josmp)

Added Paths:
-----------
    numa-bench/trunk/src/GNUmakefile

Added: numa-bench/trunk/src/GNUmakefile
===================================================================
--- numa-bench/trunk/src/GNUmakefile	                        (rev 0)
+++ numa-bench/trunk/src/GNUmakefile	2008-02-26 19:40:47 UTC (rev 510)
@@ -0,0 +1,370 @@
+OUTDIR := /home/yang/work/assorted/numa-bench/trunk/src/out
+SRCDIR := /home/yang/work/assorted/numa-bench/trunk/src
+SRCPATH := /home/yang/work/assorted/numa-bench/trunk/src
+all: malloc avail
+PREFIX_0 := /opt/malloc
+BINDIR_0 := $(PREFIX_0)/bin
+
+LIBS_0 := -lpthread
+SRCS_0 := $(SRCDIR)/malloc.cc
+ORIGTARGET_0 := malloc
+AUTOLIBS_0 :=  
+LANG_0 := cpp
+TARGET_0 := malloc-dbg
+FLAGS_0 := -g3
+
+OBJDIR_0 := $(OUTDIR)/$(TARGET_0)-obj
+CFLAGS_0 := $(CFLAGS) $(FLAGS_0)
+CXXFLAGS_0 := $(CXXFLAGS) $(FLAGS_0)
+CPPFLAGS_0 := $(CPPFLAGS)
+LDFLAGS_0 := $(LDFLAGS) $(LIBS_0) $(AUTOLIBS_0)
+EXE_0 := $(OUTDIR)/$(TARGET_0)
+
+OBJS_0 := $(foreach ext, c cc cpp cxx, \
+              $(patsubst $(SRCDIR)/%.$(ext), $(OBJDIR_0)/%.o, \
+                $(filter %.$(ext), $(SRCS_0))))
+
+CC_0 := gcc
+CXX_0 := g++
+ifeq ($(LANG_0),c)
+COMPILER_0 := $(CC_0)
+else
+COMPILER_0 := $(CXX_0)
+endif
+
+$(TARGET_0): $(EXE_0)
+
+$(EXE_0): $(OBJS_0)
+	$(COMPILER_0) $(LDFLAGS_0) -o $@ $^
+
+$(OBJDIR_0)/%.o: $(SRCDIR)/%.c
+	mkdir -p $(@D)
+	$(CC_0) $(CPPFLAGS_0) $(CFLAGS_0) -MD -c -o $@ $<
+
+$(OBJDIR_0)/%.o: $(SRCDIR)/%.cc
+	mkdir -p $(@D)
+	$(CXX_0) $(CPPFLAGS_0) $(CXXFLAGS_0) -MD -c -o $@ $<
+
+$(OBJDIR_0)/%.o: $(SRCDIR)/%.cpp
+	mkdir -p $(@D)
+	$(CXX_0) $(CPPFLAGS_0) $(CXXFLAGS_0) -MD -c -o $@ $<
+
+$(OBJDIR_0)/%.o: $(SRCDIR)/%.cxx
+	mkdir -p $(@D)
+	$(CXX_0) $(CPPFLAGS_0) $(CXXFLAGS_0) -MD -c -o $@ $<
+
+clean-$(TARGET_0):
+	rm -rf $(OBJDIR_0) $(EXE_0)
+
+$(ORIGTARGET_0): $(TARGET_0)
+
+.PHONY: $(ORIGTARGET_0)
+PREFIX_1 := /opt/malloc
+BINDIR_1 := $(PREFIX_1)/bin
+
+LIBS_1 := -lpthread
+SRCS_1 := $(SRCDIR)/malloc.cc
+ORIGTARGET_1 := malloc
+AUTOLIBS_1 :=  
+LANG_1 := cpp
+TARGET_1 := malloc-opt
+FLAGS_1 := -O3
+
+OBJDIR_1 := $(OUTDIR)/$(TARGET_1)-obj
+CFLAGS_1 := $(CFLAGS) $(FLAGS_1)
+CXXFLAGS_1 := $(CXXFLAGS) $(FLAGS_1)
+CPPFLAGS_1 := $(CPPFLAGS)
+LDFLAGS_1 := $(LDFLAGS) $(LIBS_1) $(AUTOLIBS_1)
+EXE_1 := $(OUTDIR)/$(TARGET_1)
+
+OBJS_1 := $(foreach ext, c cc cpp cxx, \
+              $(patsubst $(SRCDIR)/%.$(ext), $(OBJDIR_1)/%.o, \
+                $(filter %.$(ext), $(SRCS_1))))
+
+CC_1 := gcc
+CXX_1 := g++
+ifeq ($(LANG_1),c)
+COMPILER_1 := $(CC_1)
+else
+COMPILER_1 := $(CXX_1)
+endif
+
+$(TARGET_1): $(EXE_1)
+
+$(EXE_1): $(OBJS_1)
+	$(COMPILER_1) $(LDFLAGS_1) -o $@ $^
+
+$(OBJDIR_1)/%.o: $(SRCDIR)/%.c
+	mkdir -p $(@D)
+	$(CC_1) $(CPPFLAGS_1) $(CFLAGS_1) -MD -c -o $@ $<
+
+$(OBJDIR_1)/%.o: $(SRCDIR)/%.cc
+	mkdir -p $(@D)
+	$(CXX_1) $(CPPFLAGS_1) $(CXXFLAGS_1) -MD -c -o $@ $<
+
+$(OBJDIR_1)/%.o: $(SRCDIR)/%.cpp
+	mkdir -p $(@D)
+	$(CXX_1) $(CPPFLAGS_1) $(CXXFLAGS_1) -MD -c -o $@ $<
+
+$(OBJDIR_1)/%.o: $(SRCDIR)/%.cxx
+	mkdir -p $(@D)
+	$(CXX_1) $(CPPFLAGS_1) $(CXXFLAGS_1) -MD -c -o $@ $<
+
+clean-$(TARGET_1):
+	rm -rf $(OBJDIR_1) $(EXE_1)
+
+$(ORIGTARGET_1): $(TARGET_1)
+
+.PHONY: $(ORIGTARGET_1)
+PREFIX_2 := /opt/malloc
+BINDIR_2 := $(PREFIX_2)/bin
+
+LIBS_2 := -lpthread
+SRCS_2 := $(SRCDIR)/malloc.cc
+ORIGTARGET_2 := malloc
+AUTOLIBS_2 :=  
+LANG_2 := cpp
+TARGET_2 := malloc-gprof
+FLAGS_2 := -pg
+
+OBJDIR_2 := $(OUTDIR)/$(TARGET_2)-obj
+CFLAGS_2 := $(CFLAGS) $(FLAGS_2)
+CXXFLAGS_2 := $(CXXFLAGS) $(FLAGS_2)
+CPPFLAGS_2 := $(CPPFLAGS)
+LDFLAGS_2 := $(LDFLAGS) $(LIBS_2) $(AUTOLIBS_2)
+EXE_2 := $(OUTDIR)/$(TARGET_2)
+
+OBJS_2 := $(foreach ext, c cc cpp cxx, \
+              $(patsubst $(SRCDIR)/%.$(ext), $(OBJDIR_2)/%.o, \
+                $(filter %.$(ext), $(SRCS_2))))
+
+CC_2 := gcc
+CXX_2 := g++
+ifeq ($(LANG_2),c)
+COMPILER_2 := $(CC_2)
+else
+COMPILER_2 := $(CXX_2)
+endif
+
+$(TARGET_2): $(EXE_2)
+
+$(EXE_2): $(OBJS_2)
+	$(COMPILER_2) $(LDFLAGS_2) -o $@ $^
+
+$(OBJDIR_2)/%.o: $(SRCDIR)/%.c
+	mkdir -p $(@D)
+	$(CC_2) $(CPPFLAGS_2) $(CFLAGS_2) -MD -c -o $@ $<
+
+$(OBJDIR_2)/%.o: $(SRCDIR)/%.cc
+	mkdir -p $(@D)
+	$(CXX_2) $(CPPFLAGS_2) $(CXXFLAGS_2) -MD -c -o $@ $<
+
+$(OBJDIR_2)/%.o: $(SRCDIR)/%.cpp
+	mkdir -p $(@D)
+	$(CXX_2) $(CPPFLAGS_2) $(CXXFLAGS_2) -MD -c -o $@ $<
+
+$(OBJDIR_2)/%.o: $(SRCDIR)/%.cxx
+	mkdir -p $(@D)
+	$(CXX_2) $(CPPFLAGS_2) $(CXXFLAGS_2) -MD -c -o $@ $<
+
+clean-$(TARGET_2):
+	rm -rf $(OBJDIR_2) $(EXE_2)
+
+$(ORIGTARGET_2): $(TARGET_2)
+
+.PHONY: $(ORIGTARGET_2)
+$(OUTDIR)/.deps: $(shell find $(OUTDIR) -name '*.d' 2> /dev/null || true)
+	mkdir -p $(OUTDIR)
+	[ "$^" ] && cat $^ > $@
+
+-include $(OUTDIR)/.deps
+
+
+
+PREFIX_3 := /opt/avail
+BINDIR_3 := $(PREFIX_3)/bin
+
+LIBS_3 := -lpthread
+SRCS_3 := $(SRCDIR)/avail.cc
+ORIGTARGET_3 := avail
+AUTOLIBS_3 := -lnuma
+LANG_3 := cpp
+TARGET_3 := avail-dbg
+FLAGS_3 := -g3
+
+OBJDIR_3 := $(OUTDIR)/$(TARGET_3)-obj
+CFLAGS_3 := $(CFLAGS) $(FLAGS_3)
+CXXFLAGS_3 := $(CXXFLAGS) $(FLAGS_3)
+CPPFLAGS_3 := $(CPPFLAGS)
+LDFLAGS_3 := $(LDFLAGS) $(LIBS_3) $(AUTOLIBS_3)
+EXE_3 := $(OUTDIR)/$(TARGET_3)
+
+OBJS_3 := $(foreach ext, c cc cpp cxx, \
+              $(patsubst $(SRCDIR)/%.$(ext), $(OBJDIR_3)/%.o, \
+                $(filter %.$(ext), $(SRCS_3))))
+
+CC_3 := gcc
+CXX_3 := g++
+ifeq ($(LANG_3),c)
+COMPILER_3 := $(CC_3)
+else
+COMPILER_3 := $(CXX_3)
+endif
+
+$(TARGET_3): $(EXE_3)
+
+$(EXE_3): $(OBJS_3)
+	$(COMPILER_3) $(LDFLAGS_3) -o $@ $^
+
+$(OBJDIR_3)/%.o: $(SRCDIR)/%.c
+	mkdir -p $(@D)
+	$(CC_3) $(CPPFLAGS_3) $(CFLAGS_3) -MD -c -o $@ $<
+
+$(OBJDIR_3)/%.o: $(SRCDIR)/%.cc
+	mkdir -p $(@D)
+	$(CXX_3) $(CPPFLAGS_3) $(CXXFLAGS_3) -MD -c -o $@ $<
+
+$(OBJDIR_3)/%.o: $(SRCDIR)/%.cpp
+	mkdir -p $(@D)
+	$(CXX_3) $(CPPFLAGS_3) $(CXXFLAGS_3) -MD -c -o $@ $<
+
+$(OBJDIR_3)/%.o: $(SRCDIR)/%.cxx
+	mkdir -p $(@D)
+	$(CXX_3) $(CPPFLAGS_3) $(CXXFLAGS_3) -MD -c -o $@ $<
+
+clean-$(TARGET_3):
+	rm -rf $(OBJDIR_3) $(EXE_3)
+
+$(ORIGTARGET_3): $(TARGET_3)
+
+.PHONY: $(ORIGTARGET_3)
+PREFIX_4 := /opt/avail
+BINDIR_4 := $(PREFIX_4)/bin
+
+LIBS_4 := -lpthread
+SRCS_4 := $(SRCDIR)/avail.cc
+ORIGTARGET_4 := avail
+AUTOLIBS_4 := -lnuma
+LANG_4 := cpp
+TARGET_4 := avail-opt
+FLAGS_4 := -O3
+
+OBJDIR_4 := $(OUTDIR)/$(TARGET_4)-obj
+CFLAGS_4 := $(CFLAGS) $(FLAGS_4)
+CXXFLAGS_4 := $(CXXFLAGS) $(FLAGS_4)
+CPPFLAGS_4 := $(CPPFLAGS)
+LDFLAGS_4 := $(LDFLAGS) $(LIBS_4) $(AUTOLIBS_4)
+EXE_4 := $(OUTDIR)/$(TARGET_4)
+
+OBJS_4 := $(foreach ext, c cc cpp cxx, \
+              $(patsubst $(SRCDIR)/%.$(ext), $(OBJDIR_4)/%.o, \
+                $(filter %.$(ext), $(SRCS_4))))
+
+CC_4 := gcc
+CXX_4 := g++
+ifeq ($(LANG_4),c)
+COMPILER_4 := $(CC_4)
+else
+COMPILER_4 := $(CXX_4)
+endif
+
+$(TARGET_4): $(EXE_4)
+
+$(EXE_4): $(OBJS_4)
+	$(COMPILER_4) $(LDFLAGS_4) -o $@ $^
+
+$(OBJDIR_4)/%.o: $(SRCDIR)/%.c
+	mkdir -p $(@D)
+	$(CC_4) $(CPPFLAGS_4) $(CFLAGS_4) -MD -c -o $@ $<
+
+$(OBJDIR_4)/%.o: $(SRCDIR)/%.cc
+	mkdir -p $(@D)
+	$(CXX_4) $(CPPFLAGS_4) $(CXXFLAGS_4) -MD -c -o $@ $<
+
+$(OBJDIR_4)/%.o: $(SRCDIR)/%.cpp
+	mkdir -p $(@D)
+	$(CXX_4) $(CPPFLAGS_4) $(CXXFLAGS_4) -MD -c -o $@ $<
+
+$(OBJDIR_4)/%.o: $(SRCDIR)/%.cxx
+	mkdir -p $(@D)
+	$(CXX_4) $(CPPFLAGS_4) $(CXXFLAGS_4) -MD -c -o $@ $<
+
+clean-$(TARGET_4):
+	rm -rf $(OBJDIR_4) $(EXE_4)
+
+$(ORIGTARGET_4): $(TARGET_4)
+
+.PHONY: $(ORIGTARGET_4)
+PREFIX_5 := /opt/avail
+BINDIR_5 := $(PREFIX_5)/bin
+
+LIBS_5 := -lpthread
+SRCS_5 := $(SRCDIR)/avail.cc
+ORIGTARGET_5 := avail
+AUTOLIBS_5 := -lnuma
+LANG_5 := cpp
+TARGET_5 := avail-gprof
+FLAGS_5 := -pg
+
+OBJDIR_5 := $(OUTDIR)/$(TARGET_5)-obj
+CFLAGS_5 := $(CFLAGS) $(FLAGS_5)
+CXXFLAGS_5 := $(CXXFLAGS) $(FLAGS_5)
+CPPFLAGS_5 := $(CPPFLAGS)
+LDFLAGS_5 := $(LDFLAGS) $(LIBS_5) $(AUTOLIBS_5)
+EXE_5 := $(OUTDIR)/$(TARGET_5)
+
+OBJS_5 := $(foreach ext, c cc cpp cxx, \
+              $(patsubst $(SRCDIR)/%.$(ext), $(OBJDIR_5)/%.o, \
+                $(filter %.$(ext), $(SRCS_5))))
+
+CC_5 := gcc
+CXX_5 := g++
+ifeq ($(LANG_5),c)
+COMPILER_5 := $(CC_5)
+else
+COMPILER_5 := $(CXX_5)
+endif
+
+$(TARGET_5): $(EXE_5)
+
+$(EXE_5): $(OBJS_5)
+	$(COMPILER_5) $(LDFLAGS_5) -o $@ $^
+
+$(OBJDIR_5)/%.o: $(SRCDIR)/%.c
+	mkdir -p $(@D)
+	$(CC_5) $(CPPFLAGS_5) $(CFLAGS_5) -MD -c -o $@ $<
+
+$(OBJDIR_5)/%.o: $(SRCDIR)/%.cc
+	mkdir -p $(@D)
+	$(CXX_5) $(CPPFLAGS_5) $(CXXFLAGS_5) -MD -c -o $@ $<
+
+$(OBJDIR_5)/%.o: $(SRCDIR)/%.cpp
+	mkdir -p $(@D)
+	$(CXX_5) $(CPPFLAGS_5) $(CXXFLAGS_5) -MD -c -o $@ $<
+
+$(OBJDIR_5)/%.o: $(SRCDIR)/%.cxx
+	mkdir -p $(@D)
+	$(CXX_5) $(CPPFLAGS_5) $(CXXFLAGS_5) -MD -c -o $@ $<
+
+clean-$(TARGET_5):
+	rm -rf $(OBJDIR_5) $(EXE_5)
+
+$(ORIGTARGET_5): $(TARGET_5)
+
+.PHONY: $(ORIGTARGET_5)
+$(OUTDIR)/.deps: $(shell find $(OUTDIR) -name '*.d' 2> /dev/null || true)
+	mkdir -p $(OUTDIR)
+	[ "$^" ] && cat $^ > $@
+
+-include $(OUTDIR)/.deps
+
+
+clean: clean-$(TARGET_2) clean-$(TARGET_3) clean-$(TARGET_0) clean-$(TARGET_5) clean-$(TARGET_1) clean-$(TARGET_4)
+clean:
+	rm -rf $(OUTDIR)
+GNUmakefile: /home/yang/work/assorted/numa-bench/trunk/src/build
+	simple-build
+clean-build:
+	rm GNUmakefile
+clean-all: clean clean-build
+.PHONY: clean all clean-build clean-all 


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

[Assorted-commits] SF.net SVN: assorted: [509] numa-bench/trunk/src

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

Revision: 509
          http://assorted.svn.sourceforge.net/assorted/?rev=509&view=rev
Author:   yangzhang
Date:     2008-02-26 00:36:09 -0800 (Tue, 26 Feb 2008)

Log Message:
-----------
experimenting with (most promising) build systems

Added Paths:
-----------
    numa-bench/trunk/src/CMakeLists.txt
    numa-bench/trunk/src/Jamroot

Added: numa-bench/trunk/src/CMakeLists.txt
===================================================================
--- numa-bench/trunk/src/CMakeLists.txt	                        (rev 0)
+++ numa-bench/trunk/src/CMakeLists.txt	2008-02-26 08:36:09 UTC (rev 509)
@@ -0,0 +1,7 @@
+project (numa-bench)
+
+add_executable (malloc malloc.cc)
+target_link_libraries (malloc pthread)
+
+add_executable (avail avail.cc)
+target_link_libraries (avail numa)

Added: numa-bench/trunk/src/Jamroot
===================================================================
--- numa-bench/trunk/src/Jamroot	                        (rev 0)
+++ numa-bench/trunk/src/Jamroot	2008-02-26 08:36:09 UTC (rev 509)
@@ -0,0 +1,2 @@
+lib pthread ;
+exe malloc : malloc.cc pthread : <toolset>gcc:<cxxflags>-g3 ;


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

[Assorted-commits] SF.net SVN: assorted: [508] simple-build/trunk/src/SimpleBuild.scala

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

Revision: 508
          http://assorted.svn.sourceforge.net/assorted/?rev=508&view=rev
Author:   yangzhang
Date:     2008-02-24 23:23:08 -0800 (Sun, 24 Feb 2008)

Log Message:
-----------
added google profiler variant

Modified Paths:
--------------
    simple-build/trunk/src/SimpleBuild.scala

Modified: simple-build/trunk/src/SimpleBuild.scala
===================================================================
--- simple-build/trunk/src/SimpleBuild.scala	2008-02-25 07:21:03 UTC (rev 507)
+++ simple-build/trunk/src/SimpleBuild.scala	2008-02-25 07:23:08 UTC (rev 508)
@@ -125,7 +125,7 @@
       val lang = ext2lang(srcs(0).s.extension)
       val template = loadTemplate("cpp")
       val struct = structs(lang)
-      val frags = for ((variant,xflags) <- List(("dbg", "-g3"), ("opt", "-O3"), ("gprof", "-pg"))) yield {
+      val frags = for ((variant,xflags) <- List(("dbg", "-g3"), ("opt", "-O3"), ("pprof", "-lprofiler"), ("gprof", "-pg"))) yield {
         val hdrs = (
           for ((k,v) <- attrs.m.elements ++ List(
                           (YamlStr("origtarget"), target),


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

[Assorted-commits] SF.net SVN: assorted: [507] nano-httpd/trunk/README

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

Revision: 507
          http://assorted.svn.sourceforge.net/assorted/?rev=507&view=rev
Author:   yangzhang
Date:     2008-02-24 23:21:03 -0800 (Sun, 24 Feb 2008)

Log Message:
-----------
added a minimal readme

Modified Paths:
--------------
    nano-httpd/trunk/README

Modified: nano-httpd/trunk/README
===================================================================
--- nano-httpd/trunk/README	2008-02-25 07:20:01 UTC (rev 506)
+++ nano-httpd/trunk/README	2008-02-25 07:21:03 UTC (rev 507)
@@ -1 +1,4 @@
-janky little web server
+% Absolutely Minimal Web Server
+% Yang Zhang
+
+Nothing much to see here. Was written for a JOSMP demo.


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

[Assorted-commits] SF.net SVN: assorted: [506] scala-commons/trunk/src/Makefile

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

Revision: 506
          http://assorted.svn.sourceforge.net/assorted/?rev=506&view=rev
Author:   yangzhang
Date:     2008-02-24 23:20:01 -0800 (Sun, 24 Feb 2008)

Log Message:
-----------
removed buggy header - not what i was expecting!

Modified Paths:
--------------
    scala-commons/trunk/src/Makefile

Modified: scala-commons/trunk/src/Makefile
===================================================================
--- scala-commons/trunk/src/Makefile	2008-02-25 07:19:27 UTC (rev 505)
+++ scala-commons/trunk/src/Makefile	2008-02-25 07:20:01 UTC (rev 506)
@@ -25,7 +25,7 @@
 
 $(DOCDIR)/index.html: $(SRCS)
 	mkdir -p $(DOCDIR)
-	scaladoc -d $(DOCDIR) -linksource -header $(HTMLFRAG)/header.html -bottom $(HTMLFRAG)/google-footer.html $^
+	scaladoc -d $(DOCDIR) -linksource $^
 
 publish: doc jar web
 	cp -r $(JAR) $(DOCDIR) $(WEBDIR)


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

[Assorted-commits] SF.net SVN: assorted: [505] scala-commons/trunk/src/commons/Path.scala

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

Revision: 505
          http://assorted.svn.sourceforge.net/assorted/?rev=505&view=rev
Author:   yangzhang
Date:     2008-02-24 23:19:27 -0800 (Sun, 24 Feb 2008)

Log Message:
-----------
added basename, dirname, extension to Path

Modified Paths:
--------------
    scala-commons/trunk/src/commons/Path.scala

Modified: scala-commons/trunk/src/commons/Path.scala
===================================================================
--- scala-commons/trunk/src/commons/Path.scala	2008-02-25 07:19:11 UTC (rev 504)
+++ scala-commons/trunk/src/commons/Path.scala	2008-02-25 07:19:27 UTC (rev 505)
@@ -13,6 +13,10 @@
     def /(relpath: String) =
       if (path.last == sep) path + relpath
       else path + sep + relpath
+
+    def basename = new File(path).getName
+    def dirname = new File(path).getPath
+    def extension = basename split "\\." last
   }
 
   /**


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

[Assorted-commits] SF.net SVN: assorted: [504] scala-commons/trunk/src/commons/Io.scala

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

Revision: 504
          http://assorted.svn.sourceforge.net/assorted/?rev=504&view=rev
Author:   yangzhang
Date:     2008-02-24 23:19:11 -0800 (Sun, 24 Feb 2008)

Log Message:
-----------
fixed regression

Modified Paths:
--------------
    scala-commons/trunk/src/commons/Io.scala

Modified: scala-commons/trunk/src/commons/Io.scala
===================================================================
--- scala-commons/trunk/src/commons/Io.scala	2008-02-25 07:07:25 UTC (rev 503)
+++ scala-commons/trunk/src/commons/Io.scala	2008-02-25 07:19:11 UTC (rev 504)
@@ -151,6 +151,7 @@
         "command failed with exit status " + proc.exitValue + ": " + cmd
       )
     }
+    out
   }
 
 }


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

[Assorted-commits] SF.net SVN: assorted: [503] java-reactor/trunk/README

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

Revision: 503
          http://assorted.svn.sourceforge.net/assorted/?rev=503&view=rev
Author:   yangzhang
Date:     2008-02-24 23:07:25 -0800 (Sun, 24 Feb 2008)

Log Message:
-----------
added a readme

Added Paths:
-----------
    java-reactor/trunk/README

Added: java-reactor/trunk/README
===================================================================
--- java-reactor/trunk/README	                        (rev 0)
+++ java-reactor/trunk/README	2008-02-25 07:07:25 UTC (rev 503)
@@ -0,0 +1,52 @@
+% Java Reactor
+% Yang Zhang
+
+Overview
+--------
+
+This is a completely minimal implementation of a select-based reactor framework
+for event-based asynchronous IO programming. The reactor core supports
+scheduling of tasks (a la Twisted, asio, etc.). This was written after
+witnessing the inefficiency of various other reactor-based implementations - at
+least for the very particular environment that is PlanetLab. It also helped us
+remove the need for a separate thread for the scheduled executor tasks - these
+are run in-line.
+
+Aside from its simplicity, efficiency, and support for scheduled tasks, it also
+has no dependencies.
+
+Related Work
+------------
+
+- [Apache Mina]: we originally used this for [SRON], but it lacked efficiency
+  and scheduled tasks, and also requires at least [slf4j].
+
+- [SRON]: The project from which Java Reactor was born.
+
+[Apache Mina]: http://mina.apache.org/
+[SRON]: https://moo.cmcl.cs.cmu.edu/trac/scaleron
+[slf4j]: http://www.slf4j.org/
+
+License
+-------
+
+Java Reactor is released under the [GNU GPL3].
+
+[GNU GPL3]: http://www.gnu.org/licenses/gpl-3.0.txt
+
+Links
+-----
+
+- Web site: <http://assorted.sf.net/java-reactor/>
+- ROX NIO tutorial: <http://rox-xmlrpc.sourceforge.net/niotut/index.html>
+
+Contact
+-------
+
+Copyright (C) 2008 [Yang Zhang].  
+All rights reserved.
+
+Back to [assorted.sf.net].
+
+[Yang Zhang]: http://www.mit.edu/~y_z/
+[assorted.sf.net]: http://assorted.sourceforge.net/


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