Merge branch 'dev' into fix-stats

17 files changed, 39 insertions, 34 deletions

diff --git a/src/cuda-sim/Makefile b/src/cuda-sim/Makefile
index 01bc480..541cf8f 100644
--- a/src/cuda-sim/Makefile
+++ b/src/cuda-sim/Makefile
@@ -91,16 +91,16 @@ $(OUTPUT_DIR)/lex.ptxinfo_.o: $(OUTPUT_DIR)/lex.ptxinfo_.c $(OUTPUT_DIR)/ptxinfo
 	$(CPP) -c $(CXX_OPT) $(OUTPUT_DIR)/lex.ptxinfo_.c -o $(OUTPUT_DIR)/lex.ptxinfo_.o
 
 $(OUTPUT_DIR)/ptx.tab.c: ptx.y
-	bison --name-prefix=ptx_ -v -d ptx.y --file-prefix=$(OUTPUT_DIR)/ptx
+	bison --name-prefix=ptx_ -v -d ptx.y --file-prefix=$(OUTPUT_DIR)/ptx 2> /dev/null
 
 $(OUTPUT_DIR)/ptxinfo.tab.c: ptxinfo.y
-	bison --name-prefix=ptxinfo_ -v -d ptxinfo.y --file-prefix=$(OUTPUT_DIR)/ptxinfo
+	bison --name-prefix=ptxinfo_ -v -d ptxinfo.y --file-prefix=$(OUTPUT_DIR)/ptxinfo 2> /dev/null
 
 $(OUTPUT_DIR)/lex.ptx_.c: ptx.l
-	flex --outfile=$(OUTPUT_DIR)/lex.ptx_.c ptx.l 
+	flex --outfile=$(OUTPUT_DIR)/lex.ptx_.c ptx.l 2> /dev/null
 
 $(OUTPUT_DIR)/lex.ptxinfo_.c: ptxinfo.l
-	flex --outfile=$(OUTPUT_DIR)/lex.ptxinfo_.c ptxinfo.l 
+	flex --outfile=$(OUTPUT_DIR)/lex.ptxinfo_.c ptxinfo.l 2> /dev/null
 
 clean:
 	rm -f *~ *.o *.gcda *.gcno *.gcov libgpgpu_ptx_sim.a \
diff --git a/src/cuda-sim/cuda-sim.cc b/src/cuda-sim/cuda-sim.cc
index b063512..888cf77 100644
--- a/src/cuda-sim/cuda-sim.cc
+++ b/src/cuda-sim/cuda-sim.cc
@@ -1531,7 +1531,7 @@ void function_info::ptx_jit_config(
   std::string filename_c(filename + "_c");
   snprintf(buff, 1024, "c++filt %s > %s", get_name().c_str(),
            filename_c.c_str());
-  assert(system(buff) != NULL);
+  assert(system(buff) != 0);
   FILE *fp = fopen(filename_c.c_str(), "r");
   char * ptr = fgets(buff, 1024, fp);
   if(ptr == NULL ){
diff --git a/src/cuda-sim/instructions.cc b/src/cuda-sim/instructions.cc
index e22d88a..4981c99 100644
--- a/src/cuda-sim/instructions.cc
+++ b/src/cuda-sim/instructions.cc
@@ -1948,7 +1948,7 @@ void mma_impl(const ptx_instruction *pI, core_t *core, warp_inst_t inst) {
             hex_val = (v[k / 2].s64 & 0xffff);
           else
             hex_val = ((v[k / 2].s64 & 0xffff0000) >> 16);
-          nw_v[k].f16 = *((half *)&hex_val);
+          nw_v[k].f16 = *(reinterpret_cast<half*>(hex_val));
         }
       }
       if (!((operand_num == 3) && (type2 == F32_TYPE))) {
diff --git a/src/cuda-sim/ptx_ir.h b/src/cuda-sim/ptx_ir.h
index 8251759..7ba7171 100644
--- a/src/cuda-sim/ptx_ir.h
+++ b/src/cuda-sim/ptx_ir.h
@@ -1248,6 +1248,7 @@ class function_info {
   const ptx_version &get_ptx_version() const {
     return m_symtab->get_ptx_version();
   }
+  virtual ~function_info(){}
   unsigned get_sm_target() const { return m_symtab->get_sm_target(); }
   bool is_extern() const { return m_extern; }
   void set_name(const char *name) { m_name = name; }
diff --git a/src/gpgpu-sim/addrdec.cc b/src/gpgpu-sim/addrdec.cc
index f4f83f9..db27c82 100644
--- a/src/gpgpu-sim/addrdec.cc
+++ b/src/gpgpu-sim/addrdec.cc
@@ -584,7 +584,7 @@ unsigned next_powerOf2(unsigned n) {
   n = n - 1;
 
   // do till only one bit is left
-  while (n & n - 1) n = n & (n - 1);  // unset rightmost bit
+  while (n & (n - 1)) n = n & (n - 1);  // unset rightmost bit
 
   // n is now a power of two (less than n)
 
diff --git a/src/gpgpu-sim/dram.cc b/src/gpgpu-sim/dram.cc
index 662c2ed..53c8238 100644
--- a/src/gpgpu-sim/dram.cc
+++ b/src/gpgpu-sim/dram.cc
@@ -880,4 +880,5 @@ unsigned dram_t::get_bankgrp_number(unsigned i) {
   } else {
     assert(1);
   }
+  return 0; // we should never get here
 }
diff --git a/src/gpgpu-sim/gpu-cache.h b/src/gpgpu-sim/gpu-cache.h
index aa693b5..ad41320 100644
--- a/src/gpgpu-sim/gpu-cache.h
+++ b/src/gpgpu-sim/gpu-cache.h
@@ -499,6 +499,7 @@ struct sector_cache_block : public cache_block_t {
     for (unsigned i = 0; i < SECTOR_CHUNCK_SIZE; ++i) {
       if (sector_mask.to_ulong() & (1 << i)) return i;
     }
+    return SECTOR_CHUNCK_SIZE; //error
   }
 };
 
diff --git a/src/gpgpu-sim/gpu-sim.cc b/src/gpgpu-sim/gpu-sim.cc
index ea50fa0..47c0b4a 100644
--- a/src/gpgpu-sim/gpu-sim.cc
+++ b/src/gpgpu-sim/gpu-sim.cc
@@ -80,7 +80,7 @@ class gpgpu_sim_wrapper {};
 #include <sstream>
 #include <string>
 
-#define MAX(a, b) (((a) > (b)) ? (a) : (b))
+// #define MAX(a, b) (((a) > (b)) ? (a) : (b)) //redefined 
 
 bool g_interactive_debugger_enabled = false;
 
diff --git a/src/gpgpu-sim/local_interconnect.cc b/src/gpgpu-sim/local_interconnect.cc
index df6bd7b..fe7bc74 100644
--- a/src/gpgpu-sim/local_interconnect.cc
+++ b/src/gpgpu-sim/local_interconnect.cc
@@ -148,8 +148,8 @@ void xbar_router::RR_Advance() {
       }
     }
   }
-
-  next_node_id = (++next_node_id % total_nodes);
+  next_node_id = next_node_id + 1 ;
+  next_node_id = (next_node_id % total_nodes);
 
   conflicts += conflict_sub;
   if (active) {
diff --git a/src/gpgpu-sim/mem_fetch.cc b/src/gpgpu-sim/mem_fetch.cc
index 456d891..0d86046 100644
--- a/src/gpgpu-sim/mem_fetch.cc
+++ b/src/gpgpu-sim/mem_fetch.cc
@@ -84,10 +84,10 @@ mem_fetch::~mem_fetch() { m_status = MEM_FETCH_DELETED; }
 #undef MF_TUP_END
 
 void mem_fetch::print(FILE *fp, bool print_inst) const {
-  if (this == NULL) {
-    fprintf(fp, " <NULL mem_fetch pointer>\n");
-    return;
-  }
+  // if (this == NULL) { // doenst make sense!
+  //   fprintf(fp, " <NULL mem_fetch pointer>\n");
+  //   return;
+  // }
   fprintf(fp, "  mf: uid=%6u, sid%02u:w%02u, part=%u, ", m_request_uid, m_sid,
           m_wid, m_raw_addr.chip);
   m_access.print(fp);
diff --git a/src/gpgpu-sim/shader.cc b/src/gpgpu-sim/shader.cc
index 69992f7..229b305 100644
--- a/src/gpgpu-sim/shader.cc
+++ b/src/gpgpu-sim/shader.cc
@@ -481,7 +481,7 @@ shader_core_ctx::shader_core_ctx(class gpgpu_sim *gpu,
   m_config = config;
   m_memory_config = mem_config;
   m_stats = stats;
-  unsigned warp_size = config->warp_size;
+  // unsigned warp_size = config->warp_size;
   Issue_Prio = 0;
 
   m_sid = shader_id;
diff --git a/src/gpgpu-sim/shader.h b/src/gpgpu-sim/shader.h
index 9243077..381e2c9 100644
--- a/src/gpgpu-sim/shader.h
+++ b/src/gpgpu-sim/shader.h
@@ -351,8 +351,8 @@ class scheduler_unit {  // this can be copied freely, so can be used in std
         m_sfu_out(sfu_out),
         m_int_out(int_out),
         m_tensor_core_out(tensor_core_out),
-        m_spec_cores_out(spec_cores_out),
         m_mem_out(mem_out),
+        m_spec_cores_out(spec_cores_out),
         m_id(id) {}
   virtual ~scheduler_unit() {}
   virtual void add_supervised_warp_id(int i) {
diff --git a/src/gpgpu-sim/shader_trace.h b/src/gpgpu-sim/shader_trace.h
index e7486d8..367262c 100644
--- a/src/gpgpu-sim/shader_trace.h
+++ b/src/gpgpu-sim/shader_trace.h
@@ -38,7 +38,7 @@
 #define SCHED_PRINT_STR SHADER_PRINT_STR "Scheduler %d - "
 #define SHADER_DTRACE(x) \
   (DTRACE(x) &&          \
-   (Trace::sampling_core == get_sid() || Trace::sampling_core == -1))
+   (Trace::sampling_core == (int)get_sid() || Trace::sampling_core == -1))
 
 // Intended to be called from inside components of a shader core.
 // Depends on a get_sid() function
diff --git a/src/intersim2/Makefile b/src/intersim2/Makefile
index 3eeeb70..dad436a 100644
--- a/src/intersim2/Makefile
+++ b/src/intersim2/Makefile
@@ -136,10 +136,10 @@ depend:
 	makedepend -f$(OBJDIR)/Makefile.makedepend -I$(INCPATH) -p$(OBJDIR)/ $(ALL_SRCS) 2> /dev/null
 
 ${LEX_OBJS}: $(OBJDIR)/lex.yy.c $(OBJDIR)/y.tab.h
-	$(CC) $(CPPFLAGS) -c $< -o $@
+	$(CC) -Wno-unused-function $(CPPFLAGS) -c $< -o $@
 
 ${YACC_OBJS}: $(OBJDIR)/y.tab.c $(OBJDIR)/y.tab.h
-	$(CC) $(CPPFLAGS) -c $< -o $@
+	$(CC) -Wno-unused-function $(CPPFLAGS) -c $< -o $@
 
 ${OBJDIR}/%.o: %.cpp 
 	$(CXX) $(CPPFLAGS) -c $< -o $@
diff --git a/src/intersim2/networks/dragonfly.cpp b/src/intersim2/networks/dragonfly.cpp
index 01a2281..f5b637e 100644
--- a/src/intersim2/networks/dragonfly.cpp
+++ b/src/intersim2/networks/dragonfly.cpp
@@ -111,7 +111,7 @@ int dragonfly_port(int rID, int source, int dest){
   int dest_grp_ID = int(dest/_grp_num_nodes);
   int grp_output=-1;
   int grp_RID=-1;
-  int group_dest=-1;
+  // int group_dest=-1;
   
   //which router within this group the packet needs to go to
   if (dest_grp_ID == grp_ID) {
@@ -123,7 +123,7 @@ int dragonfly_port(int rID, int source, int dest){
       grp_output = dest_grp_ID - 1;
     }
     grp_RID = int(grp_output /gP) + grp_ID * _grp_num_routers;
-    group_dest = grp_RID * gP;
+    // group_dest = grp_RID * gP;
   }
 
   //At the last hop
@@ -221,7 +221,7 @@ void DragonFlyNew::_BuildNet( const Configuration &config )
   int _input=-1;
   int _dim_ID=-1;
   int _num_ports_per_switch=-1;
-  int _dim_size=-1;
+  // int _dim_size=-1;
   int c;
 
   ostringstream router_name;
@@ -314,7 +314,7 @@ void DragonFlyNew::_BuildNet( const Configuration &config )
     // intra-group GROUP channels
     for ( int dim = 0; dim < _n; ++dim ) {
 
-      _dim_size = powi(_k,dim);
+      // _dim_size = powi(_k,dim);
 
       _dim_ID = ((int) (node / ( powi(_p, dim))));
 
@@ -356,16 +356,16 @@ void DragonFlyNew::_BuildNet( const Configuration &config )
 
 
     // add INPUT channels -- "optical" channels connecting the groups
-    int _grp_num_routers;
+    // int _grp_num_routers;
     int grp_output;
-    int grp_ID2;
+    // int grp_ID2;
 
     for ( int cnt = 0; cnt < _p; ++cnt ) {
       //	   _dim_ID
       grp_output = _dim_ID* _p + cnt;
 
-      _grp_num_routers = powi(_k, _n-1);
-      grp_ID2 = (int) ((grp_ID - 1) / (_k - 1));
+      // _grp_num_routers = powi(_k, _n-1);
+      // grp_ID2 = (int) ((grp_ID - 1) / (_k - 1));
 
       if ( grp_ID > grp_output)   {
 
@@ -495,8 +495,8 @@ void ugal_dragonflynew( const Router *r, const Flit *f, int in_channel,
   int debug = f->watch;
   int out_port = -1;
   int out_vc = 0;
-  int min_queue_size, min_hopcnt;
-  int nonmin_queue_size, nonmin_hopcnt;
+  int min_queue_size; //, min_hopcnt;
+  int nonmin_queue_size; //, nonmin_hopcnt;
   int intm_grp_ID;
   int intm_rID;
 
@@ -523,13 +523,13 @@ void ugal_dragonflynew( const Router *r, const Flit *f, int in_channel,
 	f->ph = 1;
       } else {
 	//congestion metrics using queue length, obtained by GetUsedCredit()
-	min_hopcnt = dragonflynew_hopcnt(f->src, f->dest);
+	// min_hopcnt = dragonflynew_hopcnt(f->src, f->dest);
 	min_router_output = dragonfly_port(rID, f->src, f->dest); 
       	min_queue_size = max(r->GetUsedCredit(min_router_output), 0) ; 
 
       
-	nonmin_hopcnt = dragonflynew_hopcnt(f->src, f->intm) +
-	  dragonflynew_hopcnt(f->intm,f->dest);
+	// nonmin_hopcnt = dragonflynew_hopcnt(f->src, f->intm) +
+	//   dragonflynew_hopcnt(f->intm,f->dest);
 	nonmin_router_output = dragonfly_port(rID, f->src, f->intm);
 	nonmin_queue_size = max(r->GetUsedCredit(nonmin_router_output), 0);
 
diff --git a/src/intersim2/networks/flatfly_onchip.cpp b/src/intersim2/networks/flatfly_onchip.cpp
index fd17c1a..df43371 100644
--- a/src/intersim2/networks/flatfly_onchip.cpp
+++ b/src/intersim2/networks/flatfly_onchip.cpp
@@ -1204,7 +1204,7 @@ void ugal_pni_flatfly_onchip( const Router *r, const Flit *f, int in_channel,
 int find_distance (int src, int dest) {
   int dist = 0;
   int _dim   = gN;
-  int _dim_size;
+  // int _dim_size;
   
   int src_tmp= (int) src / gC;
   int dest_tmp = (int) dest / gC;
@@ -1212,7 +1212,7 @@ int find_distance (int src, int dest) {
   
   //  cout << " HOP CNT between  src: " << src << " dest: " << dest;
   for (int d=0;d < _dim; d++) {
-    _dim_size = powi(gK, d )*gC;
+    // _dim_size = powi(gK, d )*gC;
     //if ((int)(src / _dim_size) !=  (int)(dest / _dim_size))
     //   dist++;
     src_id = src_tmp % gK;
diff --git a/src/stream_manager.cc b/src/stream_manager.cc
index e99bf87..0ce3c6a 100644
--- a/src/stream_manager.cc
+++ b/src/stream_manager.cc
@@ -227,6 +227,8 @@ void stream_operation::print(FILE *fp) const {
     case stream_no_op:
       fprintf(fp, "no-op");
       break;
+    default:
+      break;
   }
 }