abstract_hardware_model.h:texrefAttr is different and can be searched via pointer

cuda-sim.cc: counts matches between cudaBinTextureToArray texref param (pointer)
to texref pointers in m_NameToTextureRef

2 files changed, 84 insertions, 57 deletions

diff --git a/src/abstract_hardware_model.h b/src/abstract_hardware_model.h
index c21ff6d..608a7e2 100644
--- a/src/abstract_hardware_model.h
+++ b/src/abstract_hardware_model.h
@@ -662,54 +662,54 @@ public:
 
     const struct textureReferenceAttr* get_texattr( const struct textureReference *texref ) const
     {
-        int matches = 0;
-        const struct textureReferenceAttr* t = NULL;
-        for (std::map<const struct textureReference*, const struct textureReferenceAttr*>::const_iterator kv = m_TextureRefToAttribute.begin(); kv!= m_TextureRefToAttribute.end(); kv ++){
-           const struct textureReference* tr = kv->first;
-           if (tr->normalized==texref->normalized&&
-                 tr->filterMode==texref->filterMode&&
-                 tr->addressMode[0]==texref->addressMode[0]&&
-                 tr->addressMode[1]==texref->addressMode[1]&&
-                 tr->addressMode[2]==texref->addressMode[2]&&
-                 tr->channelDesc.x==texref->channelDesc.x&&
-                 tr->channelDesc.y==texref->channelDesc.y&&
-                 tr->channelDesc.z==texref->channelDesc.z&&
-                 tr->channelDesc.w==texref->channelDesc.w&&
-                 tr->channelDesc.f==texref->channelDesc.f&&
-                 tr->sRGB==texref->sRGB&&
-                 tr->maxAnisotropy==texref->maxAnisotropy&&
-                 tr->mipmapFilterMode==texref->mipmapFilterMode&&
-                 tr->mipmapLevelBias==texref->mipmapLevelBias&&
-                 tr->minMipmapLevelClamp==texref->minMipmapLevelClamp&&
-                 tr->maxMipmapLevelClamp==texref->maxMipmapLevelClamp&&
-                 tr->__cudaReserved[0] ==texref->__cudaReserved[0]&&
-                 tr->__cudaReserved[1] ==texref->__cudaReserved[1]&&
-                 tr->__cudaReserved[2] ==texref->__cudaReserved[2]&&
-                 tr->__cudaReserved[3] ==texref->__cudaReserved[3]&&
-                 tr->__cudaReserved[4] ==texref->__cudaReserved[4]&&
-                 tr->__cudaReserved[5] ==texref->__cudaReserved[5]&&
-                 tr->__cudaReserved[6] ==texref->__cudaReserved[6]&&
-                 tr->__cudaReserved[7] ==texref->__cudaReserved[7]&&
-                 tr->__cudaReserved[8] ==texref->__cudaReserved[8]&&
-                 tr->__cudaReserved[9] ==texref->__cudaReserved[9]&&
-                 tr->__cudaReserved[10]==texref->__cudaReserved[10]&&
-                 tr->__cudaReserved[11]==texref->__cudaReserved[11]&&
-                 tr->__cudaReserved[12]==texref->__cudaReserved[12]&&
-                 tr->__cudaReserved[13]==texref->__cudaReserved[13]&&
-                 tr->__cudaReserved[14]==texref->__cudaReserved[14]){
-                 matches++;
-                 t = kv->second;
-                 //return kv->second;
-           }
-        }
-
-        printf("matches (texattr) = %d\n", matches);
-        //assert(matches==1);
-        return t;
+//        int matches = 0;
+//        const struct textureReferenceAttr* t = NULL;
+//        for (std::map<const struct textureReference*, const struct textureReferenceAttr*>::const_iterator kv = m_TextureRefToAttribute.begin(); kv!= m_TextureRefToAttribute.end(); kv ++){
+//           const struct textureReference* tr = kv->first;
+//           if (tr->normalized==texref->normalized&&
+//                 tr->filterMode==texref->filterMode&&
+//                 tr->addressMode[0]==texref->addressMode[0]&&
+//                 tr->addressMode[1]==texref->addressMode[1]&&
+//                 tr->addressMode[2]==texref->addressMode[2]&&
+//                 tr->channelDesc.x==texref->channelDesc.x&&
+//                 tr->channelDesc.y==texref->channelDesc.y&&
+//                 tr->channelDesc.z==texref->channelDesc.z&&
+//                 tr->channelDesc.w==texref->channelDesc.w&&
+//                 tr->channelDesc.f==texref->channelDesc.f&&
+//                 tr->sRGB==texref->sRGB&&
+//                 tr->maxAnisotropy==texref->maxAnisotropy&&
+//                 tr->mipmapFilterMode==texref->mipmapFilterMode&&
+//                 tr->mipmapLevelBias==texref->mipmapLevelBias&&
+//                 tr->minMipmapLevelClamp==texref->minMipmapLevelClamp&&
+//                 tr->maxMipmapLevelClamp==texref->maxMipmapLevelClamp&&
+//                 tr->__cudaReserved[0] ==texref->__cudaReserved[0]&&
+//                 tr->__cudaReserved[1] ==texref->__cudaReserved[1]&&
+//                 tr->__cudaReserved[2] ==texref->__cudaReserved[2]&&
+//                 tr->__cudaReserved[3] ==texref->__cudaReserved[3]&&
+//                 tr->__cudaReserved[4] ==texref->__cudaReserved[4]&&
+//                 tr->__cudaReserved[5] ==texref->__cudaReserved[5]&&
+//                 tr->__cudaReserved[6] ==texref->__cudaReserved[6]&&
+//                 tr->__cudaReserved[7] ==texref->__cudaReserved[7]&&
+//                 tr->__cudaReserved[8] ==texref->__cudaReserved[8]&&
+//                 tr->__cudaReserved[9] ==texref->__cudaReserved[9]&&
+//                 tr->__cudaReserved[10]==texref->__cudaReserved[10]&&
+//                 tr->__cudaReserved[11]==texref->__cudaReserved[11]&&
+//                 tr->__cudaReserved[12]==texref->__cudaReserved[12]&&
+//                 tr->__cudaReserved[13]==texref->__cudaReserved[13]&&
+//                 tr->__cudaReserved[14]==texref->__cudaReserved[14]){
+//                 matches++;
+//                 t = kv->second;
+//                 //return kv->second;
+//           }
+//        }
+//
+//        printf("matches (texattr) = %d\n", matches);
+//        //assert(matches==1);
+//        return t;
 
-//        std::map<const struct textureReference*, const struct textureReferenceAttr*>::const_iterator t=m_TextureRefToAttribute.find(texref);
-//        assert(t != m_TextureRefToAttribute.end());
-//        return t->second;
+        std::map<const struct textureReference*, const struct textureReferenceAttr*>::const_iterator t=m_TextureRefToAttribute.find(texref);
+        assert(t != m_TextureRefToAttribute.end());
+        return t->second;
     }
 
     const gpgpu_functional_sim_config &get_config() const { return m_function_model_config; }
diff --git a/src/cuda-sim/cuda-sim.cc b/src/cuda-sim/cuda-sim.cc
index 6e04ca8..39ffa63 100644
--- a/src/cuda-sim/cuda-sim.cc
+++ b/src/cuda-sim/cuda-sim.cc
@@ -26,6 +26,10 @@
 // OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
+#ifndef VERSION_EIGHT
+#define VERSION_EIGHT
+#endif
+
 #include "cuda-sim.h"
 
 #include "instructions.h"
@@ -104,6 +108,16 @@ static address_type get_converge_point(address_type pc);
 
 void gpgpu_t::gpgpu_ptx_sim_bindNameToTexture(const char* name, const struct textureReference* texref, int dim, int readmode, int ext)
 {
+
+//#ifdef VERSION_EIGHT
+//   int i;
+//   printf("%s  ", name);
+//   printf("__cudaReserved:");
+//   for (i = 0; i<15; i++){
+//      printf("  %i", texref->__cudaReserved[i]);
+//   }
+//   printf("\n");
+//#endif
    std::string texname(name);
    m_NameToTextureRef[texname] = texref;
    const textureReferenceAttr *texAttr = new textureReferenceAttr(texref, dim, (enum cudaTextureReadMode)readmode, ext); 
@@ -140,16 +154,15 @@ unsigned int intLOGB2( unsigned int v ) {
 
 void gpgpu_t::gpgpu_ptx_sim_bindTextureToArray(const struct textureReference* texref, const struct cudaArray* array)
 {
+//#ifdef VERSION_EIGHT
+//   int i;
+//   printf("__cudaReserved:");
+//   for (i = 0; i<15; i++){
+//      printf("  %i", texref->__cudaReserved[i]);
+//   }
+//   printf("\n");
+//#endif
    //   counts number of matches
-//   int                           normalized;
-//   enum cudaTextureFilterMode    filterMode;
-//   enum cudaTextureAddressMode   addressMode[3];
-//   struct cudaChannelFormatDesc  channelDesc;
-//   int                        x;
-//   int                        y;
-//   int                        z;
-//   int                        w;
-//   enum cudaChannelFormatKind f;
 //   int trMatches = 0;
 //   for (auto& kv : m_NameToTextureRef){
 //      const struct textureReference* tr = kv.second;
@@ -170,7 +183,21 @@ void gpgpu_t::gpgpu_ptx_sim_bindTextureToArray(const struct textureReference* te
 //   }
 //   printf("GPGPU-Sim PTX:   matches to texref = %d\n", trMatches);
 //   assert(trMatches==1);
-   m_TextureRefToCudaArray[texref] = array;
+   
+   //tests if texref pointer matches any pointer in m_NameToTextureRef map
+   int trMatches = 0;
+   for (auto& kv : m_NameToTextureRef){
+      const struct textureReference* tr = kv.second;
+      if (tr==texref){
+         m_TextureRefToCudaArray[tr] = array;
+         //printf("%s\n", kv.first);
+         trMatches++;
+      }
+   }
+   printf("GPGPU-Sim PTX:   matches to texref = %d\n", trMatches);
+   //assert(trMatches==1);
+
+   //m_TextureRefToCudaArray[texref] = array;
    unsigned int texel_size_bits = array->desc.w + array->desc.x + array->desc.y + array->desc.z;
    unsigned int texel_size = texel_size_bits/8;
    unsigned int Tx, Ty;