explicit template instantiations

flexflow · Nov 3, 2022 · 0bf5e99 · 0bf5e99
1 parent 7b95fe7
commit 0bf5e99
Show file tree

Hide file tree

Showing 2 changed files with 26 additions and 0 deletions.
diff --git a/src/parallel_ops/partition.cpp b/src/parallel_ops/partition.cpp
@@ -50,11 +50,35 @@ void Repartition::backward_kernel(const T *output_grad_ptr,
 
 RepartitionMeta::RepartitionMeta(FFHandler handler) : OpMeta(handler) {}
 
+// float
 template void Repartition::forward_kernel<float>(float const *input_ptr,
                                                  float *output_ptr,
                                                  size_t num_elements);
 template void Repartition::backward_kernel<float>(float const *output_grad_ptr,
                                                   float *input_grad_ptr,
                                                   size_t num_elements);
+// double
+template void Repartition::forward_kernel<double>(double const *input_ptr,
+                                                 double *output_ptr,
+                                                 size_t num_elements);
+template void Repartition::backward_kernel<double>(double const *output_grad_ptr,
+                                                  double *input_grad_ptr,
+                                                  size_t num_elements);
+
+// int
+template void Repartition::forward_kernel<int>(int const *input_ptr,
+                                                 int *output_ptr,
+                                                 size_t num_elements);
+template void Repartition::backward_kernel<int>(int const *output_grad_ptr,
+                                                  int *input_grad_ptr,
+                                                  size_t num_elements);
+
+// long
+template void Repartition::forward_kernel<long>(long const *input_ptr,
+                                                 long *output_ptr,
+                                                 size_t num_elements);
+template void Repartition::backward_kernel<long>(long const *output_grad_ptr,
+                                                  long *input_grad_ptr,
+                                                  size_t num_elements);
 
 }; // namespace FlexFlow
diff --git a/src/runtime/accessor.cc b/src/runtime/accessor.cc
@@ -316,9 +316,11 @@ GenericTensorAccessorW
 
 #define DIMFUNC(DIM)                                                           \
   template class TensorAccessorR<float, DIM>;                                  \
+  template class TensorAccessorR<double, DIM>;                                  \
   template class TensorAccessorR<int32_t, DIM>;                                \
   template class TensorAccessorR<int64_t, DIM>;                                \
   template class TensorAccessorW<float, DIM>;                                  \
+  template class TensorAccessorW<double, DIM>;                                  \
   template class TensorAccessorW<int32_t, DIM>;                                \
   template class TensorAccessorW<int64_t, DIM>;
 LEGION_FOREACH_N(DIMFUNC)