Feature/core binding report

src/allvars.h

@@ -37,6 +37,9 @@
 #include <unordered_set>
 #include <unistd.h>
 #include <vector>
+#include <sys/types.h>
+#include <sys/sysctl.h>
+#include <sched.h>
 
 #include <gsl/gsl_heapsort.h>
 #include <gsl/gsl_errno.h>

src/main.cxx

@@ -83,6 +83,8 @@ void show_version_info(int argc, char *argv[])
 #else
 	<< "no";
 #endif
+    // report binding
+    report_binding();
 }
 
 int run(int argc,char **argv)

src/proto.h

@@ -890,6 +890,21 @@ namespace vr {
 std::string GetMemUsage(const std::string &function);
 #define MEMORY_USAGE_REPORT(lvl) { if(LOG_ENABLED(lvl)) LOG(lvl) << GetMemUsage(__FUNCTION__); }
 
+/// Core binding 
+//@{
+#ifdef __APPLE__
+
+#define SYSCTL_CORE_COUNT   "machdep.cpu.core_count"
+#define CPU_SETSIZE 1024
+typedef struct cpu_set {
+  uint32_t    count;
+} cpu_set_t;
+int sched_getaffinity(pid_t pid, size_t cpu_size, cpu_set_t *cpu_set);
+#endif 
+void cpuset_to_cstr(cpu_set_t *mask, char *str);
+void report_binding();
+//@}
+
 namespace vr {
 	/// Get the basename of `filename`
 	std::string basename(const std::string &filename);

src/utilities.cxx

@@ -86,6 +86,112 @@ std::string GetMemUsage(const std::string &function)
     return memory_report.str();
 }
 
+/* Borrowed from util-linux-2.13-pre7/schedutils/taskset.c */
+#ifdef __APPLE__
+
+static inline void
+CPU_ZERO(cpu_set_t *cs) { cs->count = 0; }
+
+static inline void
+CPU_SET(int num, cpu_set_t *cs) { cs->count |= (1 << num); }
+
+static inline int
+CPU_ISSET(int num, cpu_set_t *cs) { return (cs->count & (1 << num)); }
+
+int sched_getaffinity(pid_t pid, size_t cpu_size, cpu_set_t *cpu_set)
+{
+  int32_t core_count = 0;
+  size_t  len = sizeof(core_count);
+  int ret = sysctlbyname(SYSCTL_CORE_COUNT, &core_count, &len, 0, 0);
+  if (ret) {
+    printf("error while get core count %d\n", ret);
+    return -1;
+  }
+  cpu_set->count = 0;
+  for (int i = 0; i < core_count; i++) {
+    cpu_set->count |= (1 << i);
+  }
+
+  return 0;
+}
+#endif
+
+void cpuset_to_cstr(cpu_set_t *mask, char *str)
+{
+  char *ptr = str;
+  int i, j, entry_made = 0;
+  for (i = 0; i < CPU_SETSIZE; i++) {
+    if (CPU_ISSET(i, mask)) {
+      int run = 0;
+      entry_made = 1;
+      for (j = i + 1; j < CPU_SETSIZE; j++) {
+        if (CPU_ISSET(j, mask)) run++;
+        else break;
+      }
+      if (!run)
+        sprintf(ptr, "%d ", i);
+      else if (run == 1) {
+        sprintf(ptr, "%d,%d ", i, i + 1);
+        i++;
+      } else {
+        sprintf(ptr, "%d-%d ", i, i + run);
+        i += run;
+      }
+      while (*ptr != 0) ptr++;
+    }
+  }
+  ptr -= entry_made;
+  ptr = nullptr;
+}
+
+void report_binding()
+{
+    // if there is no MPI do not report any binding
+#if !defined(USEMPI) && !defined(USEOPENMP)
+    return; 
+#endif
+#ifdef USEMPI
+    MPI_Barrier(MPI_COMM_WORLD);
+#else
+    int ThisTask=0, NProcs=1;
+#endif
+    string binding_report = "Core binding \n ";
+    cpu_set_t coremask;
+    char clbuf[7 * CPU_SETSIZE], hnbuf[64];
+    memset(clbuf, 0, sizeof(clbuf));
+    memset(hnbuf, 0, sizeof(hnbuf));
+    (void)gethostname(hnbuf, sizeof(hnbuf));
+#ifdef USEOPENMP
+    #pragma omp parallel shared (binding_report) private(coremask, clbuf) 
+#endif
+    {
+        string result;
+        (void)sched_getaffinity(0, sizeof(coremask), &coremask);
+        cpuset_to_cstr(&coremask, clbuf);
+        result = "\t On node " + string(hnbuf) + " : ";
+#ifdef USEMPI 
+        result += "MPI Rank " + to_string(ThisTask) + " : ";
+#endif
+#ifdef USEOPENMP
+        auto thread = omp_get_thread_num();
+        result +=" OMP Thread " + to_string(thread) + " : ";
+#endif
+        result += " Core affinity = " + string(clbuf) + " \n ";
+#ifdef USEOPENMP 
+        #pragma omp critical 
+#endif 
+        {
+            binding_report +=result;
+
+        }
+    }
+    LOG(info)<<binding_report;
+#ifdef USEMPI
+    MPI_Barrier(MPI_COMM_WORLD);
+#endif
+}
+
+
 #ifdef NOMASS
 void VR_NOMASS(){};
 #endif