[Bug Report] invalid signbit result

[hschoi4448]

Status: There are hardware limitations that prevent -0.0 being handled. If there are use cases that cannot be handled with the current op, we need to identify them and create new ops to handle those special cases. See comments on https://github.com/tenstorrent/tt-metal/issues/8944

Describe the bug A clear and concise description of what the bug is.

To Reproduce Steps to reproduce the behavior:

1. Copy and past below code

   
   # SPDX-FileCopyrightText: © 2023 Tenstorrent Inc.

SPDX-License-Identifier: Apache-2.0

import torch import pytest import tt_lib from tests.tt_eager.python_api_testing.unit_testing.backward_ops.utility_funcs import data_gen_pt_tt, compare_results

import ttnn from tests.tt_eager.python_api_testing.sweep_tests import ( pytorch_ops, tt_lib_ops ) from tests.ttnn.python_api_testing.sweep_tests import ttnn_ops

def data_gen_pt_tt(input_shapes, device, required_grad=False, val=1): pt_tensor = (torch.ones(input_shapes, requires_grad=required_grad) * val).bfloat16() tt_tensor = ( tt_lib.tensor.Tensor(pt_tensor, tt_lib.tensor.DataType.BFLOAT16).to(tt_lib.tensor.Layout.TILE).to(device) ) return pt_tensor, tt_tensor

@pytest.mark.parametrize( "input_shapes", ( (torch.Size([1, 1, 32, 32])), ), ) def test1(input_shapes, device): val = float('0') in_data, input_tensor = data_gen_pt_tt(input_shapes, device, True, val=val)

print("input_tensor", input_tensor)

golden_tensor = pytorch_ops.signbit(in_data)
tt_output_tensor_on_device = ttnn.signbit(input_tensor)

print("tt_output_tensor_on_device", tt_output_tensor_on_device)
print("golden_tensor", golden_tensor)

2. Run with pytest
```Python
input_tensor ttnn.Tensor([[[[ 0.00000,  0.00000,  ...,  0.00000,  0.00000],
               [ 0.00000,  0.00000,  ...,  0.00000,  0.00000],
               ...,
               [ 0.00000,  0.00000,  ...,  0.00000,  0.00000],
               [ 0.00000,  0.00000,  ...,  0.00000,  0.00000]]]], shape=Shape([1, 1, 32, 32]), dtype=DataType::BFLOAT16, layout=Layout::TILE)
tt_output_tensor_on_device ttnn.Tensor([[[[ 1.00000,  1.00000,  ...,  1.00000,  1.00000],
               [ 1.00000,  1.00000,  ...,  1.00000,  1.00000],
               ...,
               [ 1.00000,  1.00000,  ...,  1.00000,  1.00000],
               [ 1.00000,  1.00000,  ...,  1.00000,  1.00000]]]], shape=Shape([1, 1, 32, 32]), dtype=DataType::BFLOAT16, layout=Layout::TILE)
golden_tensor tensor([[[[False, False, False,  ..., False, False, False],
          [False, False, False,  ..., False, False, False],
          [False, False, False,  ..., False, False, False],
          ...,
          [False, False, False,  ..., False, False, False],
          [False, False, False,  ..., False, False, False],
          [False, False, False,  ..., False, False, False]]]])

Expected behavior A clear and concise description of what you expected to happen.

Screenshots If applicable, add screenshots to help explain your problem.

Please complete the following environment information:

• OS: [e.g. Ubuntu 20.04]
• Version of software (eg. commit db6d456dbf2cb92be52179612a7264956ff1d286)
• Device: wormhole_b0

Additional context Add any other context about the problem here.

[umadevimcw]

@tt-aho @eyonland and @jliangTT Not able to check -0.0 in both GS and WHB0. How do we proceed with this? Suggestions will be helpful for us to proceed further.

[hschoi4448]

Is this issue still ongoing? Or is it an issue that cannot be addressed by TT? @tt-aho @eyonland @jliangTT @umadevimcw @razorback3

[eyonland]

There are hardware limitations that prevent -0.0 being handled. If there are use cases that cannot be handled with the current op, we need to identify them and create new ops to handle those special cases. See comments on #8944