import unittest
import random
import torch
import torch_npu
import hypothesis
from torch_npu.testing.testcase import TestCase, run_tests
from torch_npu.testing.common_utils import SupportedDevices
class TestForeachSign(TestCase):
torch_dtypes = {
"float16": torch.float16,
"float32": torch.float32,
"bfloat16": torch.bfloat16,
"int32": torch.int32,
"int8": torch.int8,
"int64": torch.int64,
}
def assert_equal(self, cpu_outs, npu_outs):
for cpu_out, npu_out in zip(cpu_outs, npu_outs):
if cpu_out.shape != npu_out.shape:
self.fail("shape error")
if cpu_out.dtype != npu_out.dtype:
self.fail("dtype error!")
result = torch.allclose(cpu_out, npu_out.cpu(), rtol=0.001, atol=0.001)
if not result:
self.fail("result error!")
return True
def create_tensors(self, tensor_nums, dtype):
cpu_tensors = []
npu_tensors = []
for i in range(tensor_nums):
m = random.randint(1, 100)
n = random.randint(1, 100)
if dtype == "int32" or dtype == "int8" or dtype == "int64":
t = torch.randint(low=-100, high=100, size=(m, n), dtype=self.torch_dtypes.get(dtype))
else:
t = torch.randn((m, n), dtype=self.torch_dtypes.get(dtype))
cpu_tensors.append(t)
npu_tensors.append(t.npu())
return tuple(cpu_tensors), tuple(npu_tensors)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_out_float32_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "float32")
cpu_output = torch._foreach_sign(cpu_tensors)
npu_output = torch._foreach_sign(npu_tensors)
self.assertRtolEqual(cpu_output, npu_output)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_out_float16_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "float16")
cpu_output = torch._foreach_sign(cpu_tensors)
npu_output = torch._foreach_sign(npu_tensors)
self.assertRtolEqual(cpu_output, npu_output)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_out_bfloat16_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "bfloat16")
cpu_output = torch._foreach_sign(cpu_tensors)
npu_output = torch._foreach_sign(npu_tensors)
self.assert_equal(cpu_output, npu_output)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_out_int32_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "int32")
cpu_output = torch._foreach_sign(cpu_tensors)
npu_output = torch._foreach_sign(npu_tensors)
self.assertRtolEqual(cpu_output, npu_output)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_out_int32_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "int8")
cpu_output = torch._foreach_sign(cpu_tensors)
npu_output = torch._foreach_sign(npu_tensors)
self.assertRtolEqual(cpu_output, npu_output)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_out_int32_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "int64")
cpu_output = torch._foreach_sign(cpu_tensors)
npu_output = torch._foreach_sign(npu_tensors)
self.assertRtolEqual(cpu_output, npu_output)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_inplace_float32_shpae_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "float32")
torch._foreach_sign_(cpu_tensors)
torch._foreach_sign_(npu_tensors)
self.assertRtolEqual(cpu_tensors, npu_tensors)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_inplace_float16_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "float16")
torch._foreach_sign_(cpu_tensors)
torch._foreach_sign_(npu_tensors)
self.assertRtolEqual(cpu_tensors, npu_tensors)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_inplace_bfloat16_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "bfloat16")
torch._foreach_sign_(cpu_tensors)
torch._foreach_sign_(npu_tensors)
self.assert_equal(cpu_tensors, npu_tensors)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_inplace_int32_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "int32")
torch._foreach_sign_(cpu_tensors)
torch._foreach_sign_(npu_tensors)
self.assertRtolEqual(cpu_tensors, npu_tensors)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_inplace_int32_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "int8")
torch._foreach_sign_(cpu_tensors)
torch._foreach_sign_(npu_tensors)
self.assertRtolEqual(cpu_tensors, npu_tensors)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_inplace_int32_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "int64")
torch._foreach_sign_(cpu_tensors)
torch._foreach_sign_(npu_tensors)
self.assertRtolEqual(cpu_tensors, npu_tensors)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_inplace_float64_shape_tensor_num(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "float64")
torch._foreach_sign_(cpu_tensors)
torch._foreach_sign_(npu_tensors)
self.assertRtolEqual(cpu_tensors, npu_tensors)
@SupportedDevices(['Ascend910B'])
def test_foreach_sign_inplace_int32_different_xpu(self):
tensor_num_list = [20, 50]
for tensor_num in tensor_num_list:
cpu_tensors, npu_tensors = self.create_tensors(tensor_num, "int32")
torch._foreach_sign_([cpu_tensors[0], npu_tensors[0]])
self.assertRtolEqual(cpu_tensors[0], npu_tensors[0])
if __name__ == "__main__":
run_tests()