mindspore-Ecosystem
/
mindspore
21
0
Fork 27
Code Issues Pull Requests 6 Projects Releases Wiki Activity

!39856 The dynamic shape capability of the inventory of the Range operator is supported. The interfaces and data types are supplemented. 

Merge pull request !39856 from NaCN/range_data
This commit is contained in:
i-robot 2022-08-09 04:16:06 +00:00 committed by Gitee
parent 38065a1a9c 6e4c9bad20
commit fba8e11bcd
No known key found for this signature in database
GPG Key ID: 173E9B9CA92EEF8F
9 changed files with 135 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
docs/api/api_python/ops/mindspore.ops.func_range.rst
View File

@ -8,9 +8,9 @@ mindspore.ops.range
三个输入的数据类型必须相同。函数返回的Tensor的数据类型与输入数据类型保持一致。
参数：
- **start** (Tensor) - 标量Tensor，序列中的第一个数字。数据类型必须为int32或者float32。
- **limit** (Tensor) - 标量Tensor，序列中的数值上线，不包括其本身。数据类型必须为int32或者float32。
- **delta** (Tensor) - 标量Tensor，表述序列中数值的步长。数据类型必须为int32或者float32。
- **start** (Tensor) - 标量Tensor，序列中的第一个数字。数据类型必须为int32，int64，float32或者float64。
- **limit** (Tensor) - 标量Tensor，序列中的数值上线，不包括其本身。数据类型必须为int32，int64，float32或者float64。
- **delta** (Tensor) - 标量Tensor，表述序列中数值的步长。数据类型必须为int32，int64，float32或者float64。
返回：
一维Tensor，数据类型与输入数据类型一致。

30
mindspore/ccsrc/plugin/device/cpu/kernel/range_cpu_kernel.cc
View File

@ -22,6 +22,17 @@ namespace kernel {
namespace {
constexpr size_t kRangeInputsNum = 3;
constexpr size_t kRangeOutputsNum = 1;
template <typename T>
T Sign(T num) {
if (num > static_cast<T>(0.0)) {
return static_cast<T>(1.0);
} else if (num == static_cast<T>(0.0)) {
return static_cast<T>(0.0);
} else {
return static_cast<T>(-1.0);
}
}
} // namespace
void RangeCpuKernelMod::InitKernel(const CNodePtr &kernel_node) {
@ -40,6 +51,10 @@ bool RangeCpuKernelMod::Launch(const std::vector<kernel::AddressPtr> &inputs, co
LaunchKernel<int32_t>(inputs, outputs);
} else if (dtype_ == kNumberTypeFloat32) {
LaunchKernel<float>(inputs, outputs);
} else if (dtype_ == kNumberTypeInt64) {
LaunchKernel<int64_t>(inputs, outputs);
} else if (dtype_ == kNumberTypeFloat64) {
LaunchKernel<double>(inputs, outputs);
} else {
MS_LOG(EXCEPTION) << "For '" << kernel_name_ << "', the dtype of input must be int or float, but got "
<< TypeIdLabel(dtype_);
@ -66,17 +81,16 @@ void RangeCpuKernelMod::LaunchKernel(const std::vector<AddressPtr> &inputs, cons
}
auto output = reinterpret_cast<T *>(outputs[0]->addr);
size_t max_index = outputs[0]->size / sizeof(T) - 1;
size_t index = 0;
while ((delta > 0 && start < limit) || (delta < 0 && start > limit)) {
if (index > max_index) {
size_t max_size = outputs[0]->size / sizeof(T);
if (Sign(delta) * Sign(limit - start) > 0) {
output_size_ = static_cast<size_t>(std::ceil(static_cast<double>(limit - start) / static_cast<double>(delta)));
if (output_size_ > max_size) {
MS_LOG(EXCEPTION) << "For " << kernel_name_ << ", the output element number exceeds the maximum number.";
}
output[index] = start;
start += delta;
index++;
for (size_t index = 0; index < output_size_; index++, start += delta) {
output[index] = start;
}
}
output_size_ = index;
}
MS_KERNEL_FACTORY_REG(NativeCpuKernelMod, Range, RangeCpuKernelMod);

4
mindspore/ccsrc/plugin/device/gpu/kernel/arrays/range_gpu_kernel.cc
View File

@ -22,5 +22,9 @@ MS_REG_GPU_KERNEL_ONE(Range, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOu
RangeGpuKernelMod, float)
MS_REG_GPU_KERNEL_ONE(Range, KernelAttr().AddInputAttr(kNumberTypeInt32).AddOutputAttr(kNumberTypeInt32),
RangeGpuKernelMod, int)
MS_REG_GPU_KERNEL_ONE(Range, KernelAttr().AddInputAttr(kNumberTypeInt64).AddOutputAttr(kNumberTypeInt64),
RangeGpuKernelMod, int64_t)
MS_REG_GPU_KERNEL_ONE(Range, KernelAttr().AddInputAttr(kNumberTypeFloat64).AddOutputAttr(kNumberTypeFloat64),
RangeGpuKernelMod, double)
} // namespace kernel
} // namespace mindspore

6
mindspore/ccsrc/plugin/device/gpu/kernel/cuda_impl/cuda_ops/range_impl.cu
View File

@ -34,5 +34,11 @@ void CalRange(const int size, const float start, const float limit, const float
template CUDA_LIB_EXPORT void CalRange<float>(const int size, const float start, const float limit, const float delta,
const float *input, float *output, cudaStream_t cuda_stream);
template CUDA_LIB_EXPORT void CalRange<double>(const int size, const float start, const float limit, const float delta,
const double *input, double *output, cudaStream_t cuda_stream);
template CUDA_LIB_EXPORT void CalRange<int>(const int size, const float start, const float limit, const float delta,
const int *input, int *output, cudaStream_t cuda_stream);
template CUDA_LIB_EXPORT void CalRange<int64_t>(const int size, const float start, const float limit, const float delta,
const int64_t *input, int64_t *output, cudaStream_t cuda_stream);

2
mindspore/python/mindspore/ops/_op_impl/cpu/range.py
View File

@ -22,7 +22,9 @@ range_op_info = CpuRegOp("Range") \
.input(2, "delta") \
.output(0, "output", "required") \
.dtype_format(DataType.I32_Default, DataType.I32_Default, DataType.I32_Default, DataType.I32_Default) \
.dtype_format(DataType.I64_Default, DataType.I64_Default, DataType.I64_Default, DataType.I64_Default) \
.dtype_format(DataType.F32_Default, DataType.F32_Default, DataType.F32_Default, DataType.F32_Default) \
.dtype_format(DataType.F64_Default, DataType.F64_Default, DataType.F64_Default, DataType.F64_Default) \
.get_op_info()

6
mindspore/python/mindspore/ops/function/array_func.py
View File

@ -489,11 +489,11 @@ def range(start, limit, delta):
Args:
start (Tensor): A scalar Tensor. The first number in the sequence. Must have
type: int32 or float32.
type: int32 ,int64, float32 or float64.
limit (Tensor): A scalar Tensor. Upper limit of the sequence, exclusive. Must
have type: int32 or float32.
have type: int32 ,int64, float32 or float64.
delta (Tensor): A scalar Tensor. Number that increments `start`. Must have
type: int32 or float32.
type: int32 ,int64, float32 or float64.
Returns:
A 1-D Tensor, with the same type as the inputs.

2
mindspore/python/mindspore/ops/operations/array_ops.py
View File

@ -6209,7 +6209,7 @@ class Range(PrimitiveWithCheck):
validator.check("delta_shape", len(delta_shape), "", 0, Rel.EQ, self.name)
def check_dtype(self, start_dtype, limit_dtype, delta_dtype):
valid_dtypes = [mstype.int32, mstype.float32]
valid_dtypes = [mstype.int32, mstype.float32, mstype.int64, mstype.float64]
inputs = {"start": start_dtype, "limit": limit_dtype, "delta": delta_dtype}
validator.check_tensors_dtypes_same_and_valid(inputs, valid_dtypes, self.name)

30
tests/st/ops/cpu/test_range_op.py
View File

@ -88,6 +88,36 @@ def test_range_op_float():
assert np.array_equal(result.asnumpy(), expect)
@pytest.mark.level0
@pytest.mark.platform_x86_cpu
@pytest.mark.env_onecard
def test_range_op_int64():
"""
Feature: test Range op on CPU.
Description: test the Range when input is int64.
Expectation: result is right.
"""
range_op = ms.ops.Range()
result = range_op(ms.Tensor(2, ms.int64), ms.Tensor(5, ms.int64), ms.Tensor(2, ms.int64))
expect = np.array([2, 4], np.int64)
assert np.array_equal(result.asnumpy(), expect)
@pytest.mark.level0
@pytest.mark.platform_x86_cpu
@pytest.mark.env_onecard
def test_range_op_float64():
"""
Feature: test Range op on CPU.
Description: test the Range when input is float64.
Expectation: result is right.
"""
range_op = ms.ops.Range()
result = range_op(ms.Tensor(2, ms.float64), ms.Tensor(5, ms.float64), ms.Tensor(1, ms.float64))
expect = np.array([2, 3, 4], np.float64)
assert np.array_equal(result.asnumpy(), expect)
@pytest.mark.level0
@pytest.mark.platform_x86_cpu
@pytest.mark.env_onecard

64
tests/st/ops/gpu/test_range_op.py
View File

@ -65,6 +65,7 @@ def test_range_precision_end_equals_last_element():
np_expected = np.arange(-12000, -12053, -1, dtype=np.float32)
np.testing.assert_allclose(ms_out, np_expected, rtol=1e-5)
@pytest.mark.level1
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
@ -89,6 +90,7 @@ def test_range_int():
np_expected = np.array([3, -2, -7])
np.testing.assert_array_equal(ms_out, np_expected)
@pytest.mark.level1
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
@ -113,6 +115,67 @@ def test_range_float():
np_expected = np.array([1.5])
np.testing.assert_array_almost_equal(ms_out, np_expected)
@pytest.mark.level1
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
def test_range_int64():
"""
Feature: test Range op on GPU.
Description: test the Range when input is int64.
Expectation: result is right.
"""
range_net = RangeNet()
ms_out = range_net(Tensor(2, mstype.int64), Tensor(5, mstype.int64), Tensor(1, mstype.int64)).asnumpy()
np_expected = np.array([2, 3, 4])
np.testing.assert_array_equal(ms_out, np_expected)
range_net = RangeNet()
ms_out = range_net(Tensor(-24, mstype.int64), Tensor(1, mstype.int64), Tensor(4, mstype.int64)).asnumpy()
np_expected = np.array([-24, -20, -16, -12, -8, -4, 0])
np.testing.assert_array_equal(ms_out, np_expected)
range_net = RangeNet()
ms_out = range_net(Tensor(8, mstype.int64), Tensor(1, mstype.int64), Tensor(-1, mstype.int64)).asnumpy()
np_expected = np.array([8, 7, 6, 5, 4, 3, 2])
np.testing.assert_array_equal(ms_out, np_expected)
range_net = RangeNet()
ms_out = range_net(Tensor(3, mstype.int64), Tensor(-11, mstype.int64), Tensor(-5, mstype.int64)).asnumpy()
np_expected = np.array([3, -2, -7])
np.testing.assert_array_equal(ms_out, np_expected)
@pytest.mark.level1
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
def test_range_float64():
"""
Feature: test Range op on GPU.
Description: test the Range when input is float64.
Expectation: result is right.
"""
range_net = RangeNet()
ms_out = range_net(Tensor(2.3, mstype.float64), Tensor(5.5, mstype.float64), Tensor(1.2, mstype.float64)).asnumpy()
np_expected = np.array([2.3, 3.5, 4.7])
np.testing.assert_array_almost_equal(ms_out, np_expected)
range_net = RangeNet()
ms_out = range_net(Tensor(-4, mstype.float64), Tensor(-1, mstype.float64), Tensor(1.5, mstype.float64)).asnumpy()
np_expected = np.array([-4.0, -2.5])
np.testing.assert_array_almost_equal(ms_out, np_expected)
range_net = RangeNet()
ms_out = range_net(Tensor(8.0, mstype.float64), Tensor(1.0, mstype.float64), Tensor(-1.0, mstype.float64)).asnumpy()
np_expected = np.array([8.0, 7.0, 6.0, 5.0, 4.0, 3.0, 2.0])
np.testing.assert_array_almost_equal(ms_out, np_expected)
range_net = RangeNet()
ms_out = range_net(Tensor(1.5, mstype.float64), Tensor(-1, mstype.float64), Tensor(-18.9, mstype.float64)).asnumpy()
np_expected = np.array([1.5])
np.testing.assert_array_almost_equal(ms_out, np_expected)
@pytest.mark.level0
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
@ -123,6 +186,7 @@ def test_range_invalid_max_output_length():
_ = P.Range(None)
_ = P.Range('5')
@pytest.mark.level1
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard