Fix unsigned arithmetic bugs in reshape ops (#16253)

tenstorrent · Dec 24, 2024 · d30b8c2 · d30b8c2
1 parent be6ed46
commit d30b8c2
Showing 1 changed file with 10 additions and 10 deletions.
diff --git a/ttnn/cpp/ttnn/operations/data_movement/reshape_view/reshape.cpp b/ttnn/cpp/ttnn/operations/data_movement/reshape_view/reshape.cpp
@@ -68,12 +68,12 @@ ttnn::Tensor convert_tensor_to_rm_reshape_convert_back_to_orig_layout(
     //This function turns ND -> MD into 2D->MD for row major and 3D->MD for tiled using a 0 cost view
     const auto layout = tensor.get_layout();
     const auto tensor_shape = tensor.get_shape();
-    TT_FATAL((tensor_shape.rank()!=0), "can't do reshape from rank 0 tensor");
+    TT_FATAL((tensor_shape.rank() != 0), "Can't do reshape from rank 0 tensor");
     if(layout == ttnn::ROW_MAJOR_LAYOUT)
     {
         //Collapse into the second last dimension
         uint32_t second_dim = 1;
-        for (int i=0; i <tensor_shape.rank()-1; i++)
+        for (int64_t i = 0; i < static_cast<int64_t>(tensor_shape.rank()) - 1; ++i)
         {
             second_dim = second_dim * tensor_shape[i];
         }
@@ -97,7 +97,7 @@ ttnn::Tensor convert_tensor_to_rm_reshape_convert_back_to_orig_layout(
     {
         uint32_t third_dim = 1;
         //Collapse into the third last dimension
-        for (int i=0; i <tensor_shape.rank()-2; i++)
+        for (int64_t i = 0; i < static_cast<int64_t>(tensor_shape.rank()) - 2; ++i)
         {
             third_dim = third_dim * tensor_shape[i];
         }
@@ -120,7 +120,7 @@ ttnn::Tensor convert_tensor_to_rm_reshape_convert_back_to_orig_layout(
             pad_value
         );
     }
-    TT_FATAL(false, "layout is neither tile nor row major");
+    TT_FATAL(false, "Layout is neither tile nor row major");
 
 }
 
@@ -136,9 +136,9 @@ ttnn::Tensor fix_shape_and_perform_reshape_on_3D_TILE(
 {
     //This function turns a TILE 3D->MD into an equivalent 3D->3D conversion and then turns the 3D output back to MD using a 0 cost view
     //Collapse into the third last dimension
-    TT_FATAL((shape.rank()!=0), "can't do reshape to rank 0 tensor");
+    TT_FATAL((shape.rank() != 0), "Can't do reshape to rank 0 tensor");
     uint32_t third_dim = 1;
-    for (int i=0; i <shape.rank()-2; i++)
+    for (int64_t i = 0; i < static_cast<int64_t>(shape.rank()) - 2; ++i)
     {
         third_dim = third_dim * shape[i];
     }
@@ -170,10 +170,10 @@ ttnn::Tensor fix_shape_and_perform_reshape_on_2D_RM(
     )
 {
     //This function turns a RM 2D->MD into an equivalent 2D->2D conversion and then turns the 2D output back to MD using a 0 cost view
-    TT_FATAL((shape.rank()!=0), "can't do reshape to rank 0 tensor");
+    TT_FATAL((shape.rank() != 0), "Can't do reshape to rank 0 tensor");
     //Collapse into the second last dimension
     uint32_t second_dim = 1;
-    for (int i=0; i <shape.rank()-1; i++)
+    for (int64_t i = 0; i < static_cast<int64_t>(shape.rank()) - 1; ++i)
     {
         second_dim = second_dim * shape[i];
     }
@@ -241,8 +241,8 @@ ttnn::Tensor perform_reshape_on_2D_RM(
 
 ttnn::Shape tiling_reshape_corrector(const ttnn::Shape& shape, const uint32_t tile_first_dim, const uint32_t tile_second_dim) {
     //Apply the correct padding metadata to the target shape
-    auto padded = shape.with_tile_padding();
-    auto rank = shape.rank();
+    ttnn::Shape padded = shape.with_tile_padding();
+    int64_t rank = shape.rank();
     const int8_t correction_1 =(tile_first_dim - (int)padded[-1] % tile_first_dim) % tile_first_dim;
     if(rank == 1)
     {