Report possible savings from sharing strings

sjoelund · OpenModelica-Hudson · commit 4835303b4af8 · 2018-02-26T07:53:09.000+01:00
Belonging to [master]: - #2231
diff --git a/Compiler/FrontEnd/ComponentReference.mo b/Compiler/FrontEnd/ComponentReference.mo
@@ -3828,5 +3828,38 @@ algorithm
   File.write(file, "]");
 end writeSubscripts;
 
+function getConsumedMemory
+  input DAE.ComponentRef inCref;
+  output Real szIdents=0;
+  output Real szTypes=0;
+  output Real szSubs=0;
+protected
+  DAE.ComponentRef cr=inCref;
+  Boolean b = true;
+algorithm
+  while b loop
+    (b,cr) := match cr
+      case DAE.CREF_IDENT()
+        algorithm
+          szIdents := szIdents + System.getSizeOfData(cr.ident);
+          szTypes := szTypes + System.getSizeOfData(cr.identType);
+          szSubs := szSubs + System.getSizeOfData(cr.subscriptLst);
+        then (false,cr);
+      case DAE.CREF_ITER()
+        algorithm
+          szIdents := szIdents + System.getSizeOfData(cr.ident);
+          szTypes := szTypes + System.getSizeOfData(cr.identType);
+        then (false,cr);
+      case DAE.CREF_QUAL()
+        algorithm
+          szIdents := szIdents + System.getSizeOfData(cr.ident);
+          szTypes := szTypes + System.getSizeOfData(cr.identType);
+          szSubs := szSubs + System.getSizeOfData(cr.subscriptLst);
+        then (true,cr.componentRef);
+      else (false,cr);
+    end match;
+  end while;
+end getConsumedMemory;
+
 annotation(__OpenModelica_Interface="frontend");
 end ComponentReference;
diff --git a/Compiler/SimCode/SimCodeMain.mo b/Compiler/SimCode/SimCodeMain.mo
@@ -1317,10 +1317,10 @@ protected function serializeNotify<T>
   input T data;
   input String name;
 protected
-  Real sz,raw_sz;
+  Real sz,raw_sz,nonSharedStringSize;
 algorithm
-  (sz,raw_sz) := System.getSizeOfData(data);
-  Error.addMessage(Error.SERIALIZED_SIZE, {name, StringUtil.bytesToReadableUnit(sz), StringUtil.bytesToReadableUnit(raw_sz)});
+  (sz,raw_sz,nonSharedStringSize) := System.getSizeOfData(data);
+  Error.addMessage(Error.SERIALIZED_SIZE, {name, StringUtil.bytesToReadableUnit(sz), StringUtil.bytesToReadableUnit(raw_sz), StringUtil.bytesToReadableUnit(nonSharedStringSize)});
 end serializeNotify;
 
 annotation(__OpenModelica_Interface="backend");
diff --git a/Compiler/Util/Error.mo b/Compiler/Util/Error.mo
@@ -1021,7 +1021,7 @@ public constant Message UNIONTYPE_MISSING_TYPEVARS = MESSAGE(5044, TRANSLATION()
 public constant Message UNIONTYPE_WRONG_NUM_TYPEVARS = MESSAGE(5045, TRANSLATION(), ERROR(),
   Util.gettext("Uniontype %s has %s type variables, but got %s."));
 public constant Message SERIALIZED_SIZE = MESSAGE(5046, TRANSLATION(), NOTIFICATION(),
-  Util.gettext("%s uses %s of memory (%s without GC overhead)."));
+  Util.gettext("%s uses %s of memory (%s without GC overhead; %s is consumed by not performing String sharing)."));
 
 public constant Message COMPILER_ERROR = MESSAGE(5999, TRANSLATION(), ERROR(),
   Util.notrans("%s"));
diff --git a/Compiler/Util/System.mo b/Compiler/Util/System.mo
@@ -1311,7 +1311,8 @@ function getSizeOfData<T>
   input T data;
   output Real sz;
   output Real raw_sz "The size without granule overhead";
-external "C" sz=SystemImpl__getSizeOfData(data, raw_sz) annotation(Library = {"omcruntime"}, Documentation(info="<html>
+  output Real nonSharedStringSize "The size that could be saved if String sharing was enabled";
+external "C" sz=SystemImpl__getSizeOfData(data, raw_sz, nonSharedStringSize) annotation(Library = {"omcruntime"}, Documentation(info="<html>
 Counts the number of bytes that were allocated to hold the given data structure.
 Includes constant data and handles cycles.
 </html>"));
diff --git a/Compiler/runtime/systemimplmisc.cpp b/Compiler/runtime/systemimplmisc.cpp
@@ -3,8 +3,8 @@
 
 
 #include <string>
-#include <set>
 #include <stack>
+#include <unordered_map>
 
 using namespace std;
 
@@ -55,19 +55,20 @@ static inline size_t actualByteSize(size_t sz)
   return res;
 }
 #include <stdio.h>
-double SystemImpl__getSizeOfData(void *data, double *raw_size_res)
+double SystemImpl__getSizeOfData(void *data, double *raw_size_res, double *nonshared_str_res)
 {
-  size_t sz=0, raw_sz=0;
-  std::set<void*> handled;
+  size_t sz=0, raw_sz=0, nonshared_str_sz=0;
+  std::unordered_map<void*,void*> handled;
   std::stack<void*> work;
+  std::unordered_map<std::string, void*> strings;
   work.push(data);
   while (!work.empty()) {
     void *item = work.top();
     work.pop();
     if (handled.find(item) != handled.end()) {
       continue;
     }
-    handled.insert(item);
+    handled[item] = 0;
     if (MMC_IS_IMMEDIATE(item)) {
       /* Uses up zero space */
       continue;
@@ -83,8 +84,16 @@ double SystemImpl__getSizeOfData(void *data, double *raw_size_res)
       continue;
     }
     if (MMC_HDRISSTRING(hdr)) {
-      raw_sz += sizeof(void*)+MMC_STRLEN(item)+1;
-      sz += actualByteSize(sizeof(void*)+MMC_STRLEN(item)+1);
+      size_t t = sizeof(void*)+MMC_STRLEN(item)+1;
+      size_t actual = actualByteSize(t);
+      std::string s(MMC_STRINGDATA(item));
+      if (strings.find(s) != strings.end()) {
+        nonshared_str_sz += actual;
+      } else {
+        strings[s] = item;
+      }
+      raw_sz += t;
+      sz += actual;
       continue;
     }
     if (MMC_HDRISSTRUCT(hdr)) {
@@ -99,11 +108,12 @@ double SystemImpl__getSizeOfData(void *data, double *raw_size_res)
       }
       continue;
     }
-fprintf(stderr, "abort... bytes=%d num items=%d\n", sz, handled.size());
-printAny(item);
+    fprintf(stderr, "abort... bytes=%ld num items=%ld\n", sz, handled.size());
+    printAny(item);
     abort();
   }
   *raw_size_res = raw_sz;
+  *nonshared_str_res = nonshared_str_sz;
   return sz;
 }
 
