1 files changed, 15 insertions, 42 deletions

diff --git a/src/abstract_hardware_model.h b/src/abstract_hardware_model.h
index 22ef509..68cb693 100644
--- a/src/abstract_hardware_model.h
+++ b/src/abstract_hardware_model.h
@@ -173,9 +173,7 @@ enum _memory_op_t {
 #include <algorithm>
 
 #if !defined(__VECTOR_TYPES_H__)
-struct dim3 {
-   unsigned int x, y, z;
-};
+#include "vector_types.h"
 #endif
 struct dim3comp {
     bool operator() (const dim3 & a, const dim3 & b) const
@@ -197,7 +195,7 @@ void increment_x_then_y_then_z( dim3 &i, const dim3 &bound);
 #include "stream_manager.h"
 class stream_manager;
 struct CUstream_st;
-extern stream_manager * g_stream_manager;
+//extern stream_manager * g_stream_manager;
 //support for pinned memories added
 extern std::map<void *,void **> pinned_memory;
 extern std::map<void *, size_t> pinned_memory_size;
@@ -399,7 +397,7 @@ typedef std::vector<address_type> addr_vector_t;
 
 class simt_stack {
 public:
-    simt_stack( unsigned wid,  unsigned warpSize);
+    simt_stack( unsigned wid,  unsigned warpSize, class gpgpu_sim * gpu);
 
     void reset();
     void launch( address_type start_pc, const simt_mask_t &active_mask );
@@ -416,6 +414,7 @@ protected:
     unsigned m_warp_id;
     unsigned m_warp_size;
 
+
     enum stack_entry_type {
         STACK_ENTRY_TYPE_NORMAL = 0,
         STACK_ENTRY_TYPE_CALL
@@ -433,6 +432,8 @@ protected:
     };
 
     std::deque<simt_stack_entry> m_stack;
+
+    class gpgpu_sim * m_gpu;
 };
 
 #define GLOBAL_HEAP_START 0xC0000000
@@ -451,19 +452,7 @@ protected:
 
 #if !defined(__CUDA_RUNTIME_API_H__)
 
-enum cudaChannelFormatKind {
-   cudaChannelFormatKindSigned,
-   cudaChannelFormatKindUnsigned,
-   cudaChannelFormatKindFloat
-};
-
-struct cudaChannelFormatDesc {
-   int                        x;
-   int                        y;
-   int                        z;
-   int                        w;
-   enum cudaChannelFormatKind f;
-};
+#include "builtin_types.h"
 
 struct cudaArray {
    void *devPtr;
@@ -475,28 +464,6 @@ struct cudaArray {
    unsigned dimensions;
 };
 
-enum cudaTextureAddressMode {
-   cudaAddressModeWrap,
-   cudaAddressModeClamp
-};
-
-enum cudaTextureFilterMode {
-   cudaFilterModePoint,
-   cudaFilterModeLinear
-};
-
-enum cudaTextureReadMode {
-   cudaReadModeElementType,
-   cudaReadModeNormalizedFloat
-};
-
-struct textureReference {
-   int                           normalized;
-   enum cudaTextureFilterMode    filterMode;
-   enum cudaTextureAddressMode   addressMode[3];
-   struct cudaChannelFormatDesc  channelDesc;
-};
-
 #endif
 
 // Struct that record other attributes in the textureReference declaration 
@@ -571,6 +538,12 @@ public:
     int resume_CTA;
     int checkpoint_CTA_t;
     int checkpoint_insn_Y;
+
+    //Move some cycle core stats here instead of being global
+    unsigned long long  gpu_sim_cycle;
+    unsigned long long  gpu_tot_sim_cycle;
+
+
     void* gpu_malloc( size_t size );
     void* gpu_mallocarray( size_t count );
     void  gpu_memset( size_t dst_start_addr, int c, size_t count );
@@ -835,8 +808,8 @@ public:
 
 class mem_fetch_allocator {
 public:
-    virtual mem_fetch *alloc( new_addr_type addr, mem_access_type type, unsigned size, bool wr ) const = 0;
-    virtual mem_fetch *alloc( const class warp_inst_t &inst, const mem_access_t &access ) const = 0;
+    virtual mem_fetch *alloc( new_addr_type addr, mem_access_type type, unsigned size, bool wr, unsigned long long cycle ) const = 0;
+    virtual mem_fetch *alloc( const class warp_inst_t &inst, const mem_access_t &access, unsigned long long cycle ) const = 0;
 };
 
 // the maximum number of destination, source, or address uarch operands in a instruction