search

tenstorrent / tt-metal

:metal: TT-NN operator library, and TT-Metalium low level kernel programming model.
Apache License 2.0
415 stars 53 forks source link

ttnn.MaxPool2d fails for shallow_unet_528x528 and shallow_unet_528x256 unit test cases. #8498

Open keerthana-r-mcw opened 4 months ago

keerthana-r-mcw commented 4 months ago

ttnn.MaxPool2d fails with TT_ASSERT @ ./tt_metal/common/bfloat8.hpp:107: fp32_vec.size() % num_float_in_tile == 0 for shallow_unet_528x528 and shallow_unet_528x256 unit test cases.

To Reproduce Steps to reproduce the behavior:

  1. Checkout the branch keerthanar/shallow_unet_maxpool_issue_unit_test.
  2. Run the following command, pytest tests/ttnn/unit_tests/operations/test_max_pool2d.py:: test_unet_max_pool_wh_issue_case
  3. The following error will occur, E RuntimeError: TT_ASSERT @ ./tt_metal/common/bfloat8.hpp:107: fp32_vec.size() % num_float_in_tile == 0
dvartaniansTT commented 4 months ago

@keerthana-r-mcw and @saichandax please create a new branch off of this one and only include the relevant/failing unit tests.