未验证 提交 d4fb5c68 编写于 作者: Q Qi Li 提交者: GitHub

[NPU] add is_empty_op_npu, test=develop (#34234)

上级 1dfd857b
develop Ligoml-patch-1 ZHUI-patch-1 add_some_yaml_config addfile ascendrelease cherry_undefined_var delete_delete_addfile delete_disable_iterable_dataset_unittest delete_fix_retry_ci delete_fix_undefined_var delete_improve_sccache delete_paralleltest delete_prv-disable-more-cache delete_revert-34910-spinlocks_for_allocator delete_revert-35069-revert-34910-spinlocks_for_allocator delete_revert-36057-dev/read_flags_in_ut dingjiaweiww-patch-1 disable_iterable_dataset_unittest dy2static enable_eager_model_test final_state_gen_python_c final_state_intermediate fix-numpy-issue fix_concat_slice fix_npu_ci fix_op_flops fix_retry_ci fix_rnn_docs fix_tensor_type fix_undefined_var fixiscan fixiscan1 fixiscan2 fixiscan3 improve_sccache incubate/infrt inplace_addto make_flag_adding_easier move_embedding_to_phi move_histogram_to_pten move_sgd_to_phi move_slice_to_pten move_temporal_shift_to_phi move_yolo_box_to_phi npu_fix_alloc paralleltest preln_ernie prv-disable-more-cache prv-md-even-more prv-onednn-2.5 pten_tensor_refactor release/2.2 release/2.3 release/2.3-fc-ernie-fix release/2.4 revert-34406-add_copy_from_tensor revert-34910-spinlocks_for_allocator revert-35069-revert-34910-spinlocks_for_allocator revert-36057-dev/read_flags_in_ut revert-36201-refine_fast_threaded_ssa_graph_executor revert-36985-add_license revert-37318-refactor_dygraph_to_eager revert-37926-eager_coreops_500 revert-37956-revert-37727-pylayer_support_tuple revert-38100-mingdong revert-38301-allocation_rearrange_pr revert-38703-numpy_bf16_package_reupload revert-38732-remove_useless_header_in_elementwise_mul_grad revert-38959-Reduce_Grad revert-39143-adjust_empty revert-39227-move_trace_op_to_pten revert-39268-dev/remove_concat_fluid_kernel revert-40170-support_partial_grad revert-41056-revert-40727-move_some_activaion_to_phi revert-41065-revert-40993-mv_ele_floordiv_pow revert-41068-revert-40790-phi_new revert-41944-smaller_inference_api_test revert-42149-do-not-reset-default-stream-for-stream-safe-cuda-allocator revert-43155-fix_ut_tempfile revert-43882-revert-41944-smaller_inference_api_test revert-45808-phi/simplify_size_op revert-46827-deform_comment support_weight_transpose zhiqiu-patch-1 v2.4.0-rc0 v2.3.2 v2.3.1 v2.3.0 v2.3.0-rc0 v2.2.2 v2.2.1 v2.2.0 v2.2.0-rc0 v2.2.0-bak0
无相关合并请求
......@@ -240,6 +240,12 @@ class AllocatorFacadePrivate {
places.emplace_back(platform::XPUPlace(dev_id));
}
#endif
#ifdef PADDLE_WITH_ASCEND_CL
int device_count = platform::GetNPUDeviceCount();
for (int dev_id = 0; dev_id < device_count; ++dev_id) {
places.emplace_back(platform::NPUPlace(dev_id));
}
#endif
for (auto& p : places) {
zero_size_allocators_[p] = std::make_shared<ZeroSizeAllocator>(p);
......
/* Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the "License");
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 Licnse. */
#include "paddle/fluid/operators/is_empty_op.h"
namespace ops = paddle::operators;
namespace plat = paddle::platform;
REGISTER_OP_NPU_KERNEL(
is_empty, ops::IsEmptyOpKernel<plat::NPUDeviceContext, float>,
ops::IsEmptyOpKernel<plat::NPUDeviceContext, plat::float16>);
......@@ -192,6 +192,10 @@ void NPUMemcpySync(void *dst, const void *src, size_t count,
dst_max_count = dst_max_count ? dst_max_count : count;
VLOG(4) << dst << " " << dst_max_count << " " << src << " " << count << " "
<< kind;
if (dst == nullptr && dst_max_count == 0) {
VLOG(4) << "Dot not call aclrtMemcpy for zero_size_allocation on NPU";
return;
}
PADDLE_ENFORCE_NPU_SUCCESS(aclrtMemcpy(dst, dst_max_count, src, count, kind));
}
......
# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# 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.
from __future__ import print_function
import unittest
import numpy as np
import sys
sys.path.append("..")
from op_test import OpTest
import paddle
paddle.enable_static()
@unittest.skipIf(not paddle.is_compiled_with_npu(),
"core is not compiled with NPU")
class TestEmpty(OpTest):
def setUp(self):
self.set_npu()
self.init_dtype()
self.op_type = "is_empty"
self.set_data()
def set_npu(self):
self.__class__.use_npu = True
self.place = paddle.NPUPlace(0)
def init_dtype(self):
self.dtype = np.float32
def set_data(self):
self.inputs = {'X': np.array([1, 2, 3]).astype(self.dtype)}
self.outputs = {'Out': np.array([False])}
def test_check_output(self):
self.check_output_with_place(self.place)
@unittest.skipIf(not paddle.is_compiled_with_npu(),
"core is not compiled with NPU")
class TestNotEmpty(TestEmpty):
def set_data(self):
self.inputs = {'X': np.array([])}
self.outputs = {'Out': np.array([True])}
@unittest.skipIf(not paddle.is_compiled_with_npu(),
"core is not compiled with NPU")
class TestIsEmptyOpError(unittest.TestCase):
def test_errors(self):
paddle.enable_static()
with paddle.static.program_guard(paddle.static.Program(),
paddle.static.Program()):
input_data = np.random.random((3, 2)).astype("float32")
def test_Variable():
# the input type must be Variable
paddle.is_empty(x=input_data)
self.assertRaises(TypeError, test_Variable)
def test_type():
# dtype must be float32, float16 in NPU
x3 = paddle.static.data(
name="x3", shape=[4, 32, 32], dtype="bool")
res = paddle.is_empty(x=x3)
self.assertRaises(TypeError, test_type)
def test_name_type():
# name type must be string.
x4 = paddle.static.data(
name="x4", shape=[3, 2], dtype="float32")
res = paddle.is_empty(x=x4, name=1)
self.assertRaises(TypeError, test_name_type)
@unittest.skipIf(not paddle.is_compiled_with_npu(),
"core is not compiled with NPU")
class TestIsEmptyOpDygraph(unittest.TestCase):
def test_dygraph(self):
paddle.disable_static(paddle.NPUPlace(0))
input = paddle.rand(shape=[4, 32, 32], dtype='float32')
res = paddle.is_empty(x=input)
if __name__ == "__main__":
unittest.main()
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册
反馈
建议
客服 返回
顶部