Throw error and skip tests on ROCm systems

ci_test/unit_tests/test_unit_layer_upsample.py

@@ -280,5 +280,5 @@ def func(cluster, dirname, weekly):
 # Create test functions that can interact with PyTest
 # Note: Create test name by removing ".py" from file name
 _test_name = os.path.splitext(os.path.basename(current_file))[0]
-for _test_func in tools.create_tests(setup_experiment, _test_name, skip_clusters=["catalyst"]):
+for _test_func in tools.create_tests(setup_experiment, _test_name, skip_clusters=["tioga", "corona"]):
     globals()[_test_func.__name__] = augment_test_func(_test_func)

ci_test/unit_tests/test_unit_layer_upsample_distconv.py

@@ -355,6 +355,6 @@ def func(cluster, dirname, weekly):
 # Note: Create test name by removing ".py" from file name
 _test_name = os.path.splitext(os.path.basename(current_file))[0]
 for _test_func in tools.create_tests(setup_experiment, _test_name,
-                                     skip_clusters=["catalyst"],
+                                     skip_clusters=["tioga", "corona"],
                                      environment=environment):
     globals()[_test_func.__name__] = augment_test_func(_test_func)

include/lbann/utils/dnn_lib/miopen/upsample.hpp

@@ -0,0 +1,115 @@
+////////////////////////////////////////////////////////////////////////////////
+// Copyright (c) 2014-2024, Lawrence Livermore National Security, LLC.
+// Produced at the Lawrence Livermore National Laboratory.
+// Written by the LBANN Research Team (B. Van Essen, et al.) listed in
+// the CONTRIBUTORS file. <[email protected]>
+//
+// LLNL-CODE-697807.
+// All rights reserved.
+//
+// This file is part of LBANN: Livermore Big Artificial Neural Network
+// Toolkit. For details, see http://software.llnl.gov/LBANN or
+// https://github.com/LLNL/LBANN.
+//
+// Licensed under the Apache License, Version 2.0 (the "Licensee"); you
+// may not use this file except in compliance with the License.  You may
+// obtain a copy of the License at:
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+// implied. See the License for the specific language governing
+// permissions and limitations under the license.
+////////////////////////////////////////////////////////////////////////////////
+#ifndef LBANN_UTILS_DNN_LIB_MIOPEN_UPSAMPLE_HPP_
+#define LBANN_UTILS_DNN_LIB_MIOPEN_UPSAMPLE_HPP_
+
+#include "lbann/utils/dnn_enums.hpp"
+#include "lbann/utils/dnn_lib/helpers.hpp"
+#include "lbann/utils/gpu/helpers.hpp"
+
+#include "utils.hpp"
+
+namespace lbann {
+
+#ifdef LBANN_HAS_MIOPEN
+namespace dnn_lib {
+
+using namespace miopen;
+
+template <typename TensorDataType, typename ScalarParameterType>
+void upsample_nearest_forward(PoolingDescriptor const& poolingDesc,
+                              ScalarParameterType const& alpha_in,
+                              TensorDescriptor const& xDesc,
+                              TensorDataType const* x,
+                              ScalarParameterType const& beta_in,
+                              TensorDescriptor const& yDesc,
+                              TensorDataType* y,
+                              dnnHandle_t handle)
+{
+  LBANN_ERROR("Upsample layer is not currently supported on ROCm devices.");
+}
+
+template <typename TensorDataType, typename ScalarParameterType>
+void upsample_nearest_forward(PoolingDescriptor const& poolingDesc,
+                              ScalarParameterType const& alpha_in,
+                              TensorDescriptor const& xDesc,
+                              El::AbstractMatrix<TensorDataType> const& x,
+                              ScalarParameterType const& beta_in,
+                              TensorDescriptor const& yDesc,
+                              El::AbstractMatrix<TensorDataType>& y)
+{
+  auto multisync =
+    El::MakeMultiSync(gpu::get_sync_info(y), gpu::get_sync_info(x));
+  auto handle_manager = internal::make_default_handle_manager(multisync);
+  upsample_nearest_forward(poolingDesc,
+                           alpha_in,
+                           xDesc,
+                           x.LockedBuffer(),
+                           beta_in,
+                           yDesc,
+                           y.Buffer(),
+                           handle_manager.get());
+}
+
+template <typename TensorDataType, typename ScalarParameterType>
+void upsample_nearest_backward(PoolingDescriptor const& poolingDesc,
+                               ScalarParameterType const& alpha_in,
+                               TensorDescriptor const& dyDesc,
+                               TensorDataType const* dy,
+                               ScalarParameterType const& beta_in,
+                               TensorDescriptor const& dxDesc,
+                               TensorDataType* dx,
+                               dnnHandle_t handle)
+{
+  LBANN_ERROR("Upsample layer is not currently supported on ROCm devices.");
+}
+
+template <typename TensorDataType, typename ScalarParameterType>
+void upsample_nearest_backward(PoolingDescriptor const& poolingDesc,
+                               ScalarParameterType const& alpha_in,
+                               TensorDescriptor const& dyDesc,
+                               El::AbstractMatrix<TensorDataType> const& dy,
+                               ScalarParameterType const& beta_in,
+                               TensorDescriptor const& dxDesc,
+                               El::AbstractMatrix<TensorDataType>& dx)
+{
+  auto multisync = El::MakeMultiSync(gpu::get_sync_info(dx),
+                                     gpu::get_sync_info(dy));
+  auto handle_manager = internal::make_default_handle_manager(multisync);
+  upsample_nearest_backward(poolingDesc,
+                            alpha_in,
+                            dyDesc,
+                            dy.LockedBuffer(),
+                            beta_in,
+                            dxDesc,
+                            dx.Buffer(),
+                            handle_manager.get());
+}
+
+} // namespace dnn_lib
+#endif // LBANN_HAS_MIOPEN
+} // namespace lbann
+#endif // LBANN_UTILS_DNN_LIB_MIOPEN_UPSAMPLE_HPP_