Merge pull request plumed#1307 from Iximiel/nit/longlonglog-dict

A faster parser

Author: gtribello

CHANGES/v2.11.md

@@ -2,7 +2,10 @@
 
 This page contains changes that will end up in 2.11
 
-__Changes relevant for users:__
+### Changes relevant for users:
 
   - When using [MOLINFO](MOLINFO.md) with the `WHOLE` flag, PBCs in the following actions will be reconstructed using a minimum spanning tree based on the coordinates stored in the MOLFILE reference pdb.
 
+### Changes relevant for developers:
+
+  - The inputlines now are parsed using a dictionary instead of an array. This speeds up the parsing of massive input lines. It is not possible anymore to register lowercase keywords and the developer must always specify the keywords completely as uppercase while parsing in the action constructor. There are no differences for the user, apart the entry in the manual for the DEBUG action: the user can still specify lowercase and mixed case keywords in the input file.

configure

@@ -754,6 +754,7 @@ enable_af_cuda
 enable_af_cpu
 enable_libtorch
 enable_libmetatomic
+enable_linuxperf
 enable_openacc
 enable_openmp
 '
@@ -1459,6 +1460,8 @@ Optional Features:
   --enable-af_cpu         enable search for arrayfire_cpu, default: no
   --enable-libtorch       enable search for libtorch, default: no
   --enable-libmetatomic   enable search for libmetatomic, default: no
+  --enable-linuxperf      enable activate the compile options for a more
+                          accurate report with perf on linux, default: no
   --enable-openacc        enable search for openacc (use PLUMED_ACC_TYPE and
                           PLUMED_ACC_GPU for controlling the settings),
                           default: no
@@ -3254,6 +3257,24 @@ fi
 
 
 
+linuxperf=
+# Check whether --enable-linuxperf was given.
+if test "${enable_linuxperf+set}" = set; then :
+  enableval=$enable_linuxperf; case "${enableval}" in
+             (yes) linuxperf=true ;;
+             (no)  linuxperf=false ;;
+             (*)   as_fn_error $? "wrong argument to --enable-linuxperf" "$LINENO" 5 ;;
+  esac
+else
+  case "no" in
+             (yes) linuxperf=true ;;
+             (no)  linuxperf=false ;;
+  esac
+
+fi
+
+
+
 openacc=
 # Check whether --enable-openacc was given.
 if test "${enable_openacc+set}" = set; then :
@@ -3287,7 +3308,6 @@ if test -z "$CXXFLAGS"
 then
   CXXFLAGS=-O3
 fi
-
 # this is a list of common compilers
 compilers="g++ c++ cxx icpc"
 c_compilers="gcc cc icc"
@@ -5215,6 +5235,103 @@ rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 
 fi
 
+# useful for having a clearer (and correct) flamegraph in linux
+if test $linuxperf = true
+then
+
+  save_CXXFLAGS="$CXXFLAGS"
+  CXXFLAGS="$CXXFLAGS -fno-omit-frame-pointer"
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -fno-omit-frame-pointer" >&5
+$as_echo_n "checking whether $CXX accepts -fno-omit-frame-pointer... " >&6; }
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_link "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: not linking" >&5
+$as_echo "not linking" >&6; }; CXXFLAGS="$save_CXXFLAGS"
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }; CXXFLAGS="$save_CXXFLAGS"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+
+  save_CXXFLAGS="$CXXFLAGS"
+  CXXFLAGS="$CXXFLAGS -gdwarf"
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -gdwarf" >&5
+$as_echo_n "checking whether $CXX accepts -gdwarf... " >&6; }
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_link "$LINENO"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: not linking" >&5
+$as_echo "not linking" >&6; }; CXXFLAGS="$save_CXXFLAGS"
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }; CXXFLAGS="$save_CXXFLAGS"
+
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+fi
+
 { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX declares c++17 support" >&5
 $as_echo_n "checking whether $CXX declares c++17 support... " >&6; }
 cat confdefs.h - <<_ACEOF >conftest.$ac_ext

configure.ac

@@ -324,6 +324,7 @@ PLUMED_CONFIG_ENABLE([af_cuda],[search for arrayfire_cuda],[no])
 PLUMED_CONFIG_ENABLE([af_cpu],[search for arrayfire_cpu],[no])
 PLUMED_CONFIG_ENABLE([libtorch],[search for libtorch],[no]) #added by luigibonati
 PLUMED_CONFIG_ENABLE([libmetatomic],[search for libmetatomic],[no])
+PLUMED_CONFIG_ENABLE([linuxperf],[activate the compile options for a more accurate report with perf on linux],[no])
 PLUMED_CONFIG_ENABLE([openacc],[search for openacc (use PLUMED_ACC_TYPE and PLUMED_ACC_GPU for controlling the settings)],[no])
 
 AC_ARG_VAR(SOEXT,[extension of dynamic libraries (so/dylib)])
@@ -341,7 +342,6 @@ if test -z "$CXXFLAGS"
 then
   CXXFLAGS=-O3
 fi
-
 # this is a list of common compilers
 compilers="g++ c++ cxx icpc"
 c_compilers="gcc cc icc"
@@ -433,6 +433,13 @@ then
   PLUMED_CHECK_CXXFLAG([-Wold-style-cast])
 fi
 
+# useful for having a clearer (and correct) flamegraph in linux
+if test $linuxperf = true
+then
+  PLUMED_CHECK_CXXFLAG([-fno-omit-frame-pointer])
+  PLUMED_CHECK_CXXFLAG([-gdwarf])
+fi
+
 AC_MSG_CHECKING([whether $CXX declares c++17 support])
 AC_COMPILE_IFELSE([AC_LANG_PROGRAM([
 #if __cplusplus < 201703L

regtest/basic/rt-make-FSUM/Makefile

@@ -0,0 +1 @@
+include ../../scripts/test.make

regtest/basic/rt-make-FSUM/config

@@ -0,0 +1 @@
+type=make

regtest/basic/rt-make-FSUM/main.cpp

@@ -0,0 +1,53 @@
+#include "plumed/tools/Tools.h"
+#include <iostream>
+#include <fstream>
+#include <iomanip>
+#include <string>
+using namespace PLMD;
+
+class tee {
+//TODO:: put this tee into a common test utilities lib
+  std::ofstream ofs;
+public:
+  tee(std::string filename) : ofs(filename) {}
+  template<typename T>
+  tee& operator<<(const T& t) {
+    ofs<<t;
+    std::cout <<t;
+    return *this;
+  }
+};
+
+template<typename T>
+using FSUM=PLMD::Tools::FastStringUnorderedMap<T>;
+
+int main() {
+  try {
+    tee out("output");
+    FSUM<int> mymap;
+    //insertions
+    std::string addr;
+    for (int i=0; i<90; i+=10) {
+      addr=std::to_string(i);
+      mymap[addr]= i;
+    }
+    {
+      //This tests that conv will actually insert a  "90\0"
+      //and not a"900" with not null char at the end
+      std::string tmp = "90123456";
+      std::string_view k {tmp.c_str(),2};
+      mymap[k] = 90;
+    }
+    for (const auto & x: {
+           "90", "0", "10","20","30","40","50","60","70","80"
+         } ) {
+      auto kk = mymap.find(x);
+      plumed_assert(kk != mymap.end())
+          <<"\""<< x << "\" should have been inserted";
+      out <<std::left<< std::setw(3) << x << mymap[x] <<"\n";
+    }
+  } catch(PLMD::Exception &e) {
+    std::cerr << "Exception:" << e.what() << "\n";
+  }
+  return 0;
+}

regtest/basic/rt-make-FSUM/output.reference

@@ -0,0 +1,10 @@
+90 90
+0  0
+10 10
+20 20
+30 30
+40 40
+50 50
+60 60
+70 70
+80 80

regtest/basic/rt-make-getwords/Makefile

@@ -0,0 +1 @@
+include ../../scripts/test.make

regtest/basic/rt-make-getwords/config

@@ -0,0 +1 @@
+type=make

regtest/basic/rt-make-getwords/main.cpp

@@ -0,0 +1,89 @@
+#include "plumed/tools/Tools.h"
+#include <fstream>
+#include <iomanip>
+#include <iostream>
+
+using namespace PLMD;
+
+class tee {
+//TODO:: put this tee into a common test utilities lib
+  std::ofstream ofs;
+public:
+  tee(std::string filename) : ofs(filename) {}
+  template<typename T>
+  tee& operator<<(const T& t) {
+    ofs<<t;
+    std::cout <<t;
+    return *this;
+  }
+};
+
+int main() {
+  try {
+//NOTE:: We are simply checking that gw has the expected behaviour
+    tee out("output");
+    for (auto test : {
+           "{abcd},{a b d}","a,b,c,d","{1,2,3},4,5,{6}",
+           "@ndx:{file1 second}","@ndx:file,@ndx:{file 1 second}",
+           "@replicas:{1,2,3} {0,0,0} {2,4,6}",
+           "{1 @replicas:{2 0 4} 3}"
+         }) {
+      out << std::quoted(test)<< '\n';
+      gch::small_vector<std::string_view> gws ;
+      Tools::getWordsSimple(gws,test,", \t\n");
+      auto gw  = Tools::getWords(test,", \t\n");
+      for(auto x : gws ) {
+
+        out <<"\tgws:"<< std::quoted(x) << "\n";
+      }
+      for(auto x : gw ) {
+        out <<"\tgw :"<< std::quoted(x) << "\n";
+      }
+      out <<'\n';
+    }
+    out << "\n\nunravelReplicas\n";
+    for(auto test: {
+          "@replicas:{1,2,3} {0,0,0} {2,4,6}",
+          "@replicas:{{1,2,3} {0,0,0} {2,4,6}}",
+          "@replicas:1 2 3"
+        }) {
+      out << std::quoted(test) << "\n";
+      for (int rep : {
+             0,1,2
+           }) {
+        auto res=Tools::unravelReplicas(test, rep);
+        out << "Replica: "<< rep << "\n";
+        out << "\t" << res << " ";
+        out << "\n";
+      }
+    }
+    out << "\n\nreadVector\n";
+    for(auto test: {
+          //this will be passed as intended by the LineToken class, the output of this first line is wrong!
+          //"@replicas:{{1,2,3} {0,0,0} {2,4,6}}",
+          //"@replicas:1 2 3",
+          "1,@replicas:{1,2,3},31,999,3",
+          "1,999,3",
+          //this works as intended, sadly it is not a possible input, currently
+          "@replicas:{1,2,3},999,3",
+          "@replicas:{1,2,3},@replicas:{3,5,6},3"
+        }) {
+      out << std::quoted(test) << "\n";
+      for (int rep : {
+             0,1,2
+           }) {
+        std::vector<int> res;
+        Tools::parseVector(test,res,rep);
+        out << "Replica: "<< rep << "\n";
+        out << "\t" ;
+        for (auto x:res) {
+          out << x << " ";
+        }
+        out << "\n";
+      }
+    }
+  } catch(PLMD::Exception &e) {
+    std::cerr << "Exception:" << e.what() << "\n";
+  }
+  return 0;
+}