reverted cuda8 changes (extra fields) and assert texture bug fix's assumption

2 files changed, 45 insertions, 148 deletions

diff --git a/src/abstract_hardware_model.h b/src/abstract_hardware_model.h
index 608a7e2..412c0a8 100644
--- a/src/abstract_hardware_model.h
+++ b/src/abstract_hardware_model.h
@@ -451,33 +451,32 @@ struct textureReference {
    enum cudaTextureAddressMode   addressMode[3];
    struct cudaChannelFormatDesc  channelDesc;
 
-#ifdef VERSION_EIGHT
-    /**                                                                            
-     * Perform sRGB->linear conversion during texture read
-     */
-    int                          sRGB;
-    /** 
-     * Limit to the anisotropy ratio
-     */
-    unsigned int                 maxAnisotropy;
-    /**
-     * Mipmap filter mode
-     */
-    enum cudaTextureFilterMode   mipmapFilterMode;
-    /**
-     * Offset applied to the supplied mipmap level
-     */
-    float                        mipmapLevelBias;
-    /** 
-     * Lower end of the mipmap level range to clamp access to
-     */
-    float                        minMipmapLevelClamp;
-    /** 
-     * Upper end of the mipmap level range to clamp access to
-     */
-    float                        maxMipmapLevelClamp;
-    int                          __cudaReserved[15];
-#endif
+//following commented section applies only to CUDA_VERSION 8+
+//    /**                                                                            
+//     * Perform sRGB->linear conversion during texture read
+//     */
+//    int                          sRGB;
+//    /** 
+//     * Limit to the anisotropy ratio
+//     */
+//    unsigned int                 maxAnisotropy;
+//    /**
+//     * Mipmap filter mode
+//     */
+//    enum cudaTextureFilterMode   mipmapFilterMode;
+//    /**
+//     * Offset applied to the supplied mipmap level
+//     */
+//    float                        mipmapLevelBias;
+//    /** 
+//     * Lower end of the mipmap level range to clamp access to
+//     */
+//    float                        minMipmapLevelClamp;
+//    /** 
+//     * Upper end of the mipmap level range to clamp access to
+//     */
+//    float                        maxMipmapLevelClamp;
+//    int                          __cudaReserved[15];
 
 };
 
@@ -569,7 +568,10 @@ public:
                  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->channelDesc.f==texref->channelDesc.f 
+                 //following commented section applies only to CUDA_VERSION 8+
+                 /*
+                 &&
                  tr->sRGB==texref->sRGB&&
                  tr->maxAnisotropy==texref->maxAnisotropy&&
                  tr->mipmapFilterMode==texref->mipmapFilterMode&&
@@ -591,6 +593,7 @@ public:
                  tr->__cudaReserved[12]==texref->__cudaReserved[12]&&
                  tr->__cudaReserved[13]==texref->__cudaReserved[13]&&
                  tr->__cudaReserved[14]==texref->__cudaReserved[14]
+                 */
                  ){
                  matches++;
                  t = kv->second;
@@ -598,15 +601,9 @@ public:
            }
         }
 
-        printf("matches (texarray) = %d\n", matches);
-        //assert(matches==1);
+        //printf("matches (texarray) = %d\n", matches);
+        assert(matches==1);
         return t;
-
-        //assert(false);
-
-//        std::map<const struct textureReference*,const struct cudaArray*>::const_iterator t=m_TextureRefToCudaArray.find(texref);
-//        assert(t != m_TextureRefToCudaArray.end());
-//        return t->second;
     }
     const struct textureInfo* get_texinfo( const struct textureReference *texref ) const
     {
@@ -623,7 +620,10 @@ public:
                  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->channelDesc.f==texref->channelDesc.f
+                 //following commented section applies only to CUDA_VERSION 8+
+                 /*
+                 &&
                  tr->sRGB==texref->sRGB&&
                  tr->maxAnisotropy==texref->maxAnisotropy&&
                  tr->mipmapFilterMode==texref->mipmapFilterMode&&
@@ -644,69 +644,23 @@ public:
                  tr->__cudaReserved[11]==texref->__cudaReserved[11]&&
                  tr->__cudaReserved[12]==texref->__cudaReserved[12]&&
                  tr->__cudaReserved[13]==texref->__cudaReserved[13]&&
-                 tr->__cudaReserved[14]==texref->__cudaReserved[14]){
+                 tr->__cudaReserved[14]==texref->__cudaReserved[14]
+                 */
+                 ){
                  matches++;
                  t = kv->second;
-                 //return kv->second;
            }
         }
 
-        printf("matches (texinfo) = %d\n", matches);
-        //assert(matches==1);
+        //printf("matches (texinfo) = %d\n", matches);
+        assert(matches==1);
         return t;
-        
-//        std::map<const struct textureReference*, const struct textureInfo*>::const_iterator t=m_TextureRefToTexureInfo.find(texref);
-//        assert(t != m_TextureRefToTexureInfo.end());
-//        return t->second;
     }
 
     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;
-
+       //note textureReferenceAttr map behaves differently from cudaArray and
+       //textureInfo maps
         std::map<const struct textureReference*, const struct textureReferenceAttr*>::const_iterator t=m_TextureRefToAttribute.find(texref);
         assert(t != m_TextureRefToAttribute.end());
         return t->second;
diff --git a/src/cuda-sim/cuda-sim.cc b/src/cuda-sim/cuda-sim.cc
index 656091c..946043a 100644
--- a/src/cuda-sim/cuda-sim.cc
+++ b/src/cuda-sim/cuda-sim.cc
@@ -26,10 +26,6 @@
 // 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"
@@ -108,16 +104,6 @@ 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); 
@@ -154,50 +140,7 @@ 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 trMatches = 0;
-//   for (auto& kv : m_NameToTextureRef){
-//      const struct textureReference* tr = kv.second;
-//      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){
-//
-//         m_TextureRefToCudaArray[tr] = array;
-//         trMatches++;
-//      }
-//   }
-//   printf("GPGPU-Sim PTX:   matches to texref = %d\n", trMatches);
-//   assert(trMatches==1);
-   
-   //tests if texref pointer matches any pointer in m_NameToTextureRef map
-   int trMatches = 0;
-        for (std::map<std::string,const struct textureReference*>::const_iterator kv = m_NameToTextureRef.begin(); kv!= m_NameToTextureRef.end(); kv ++){
-      const struct textureReference* tr = kv->second;
-      if (tr==texref){
-         m_TextureRefToCudaArray[tr] = array;
-         printf("%s\n", kv->first.c_str());
-         trMatches++;
-      }
-   }
-   printf("GPGPU-Sim PTX:   matches to texref = %d\n", trMatches);
-   //assert(trMatches==1);
-
-   //m_TextureRefToCudaArray[texref] = array;
+   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;