diff --git a/test/3x/torch/algorithms/fp8_quant/unit_tests/test_layers/test_linear.py b/test/3x/torch/algorithms/fp8_quant/unit_tests/test_layers/test_linear.py index a82fa9cfade..6cc918dc1cf 100644 --- a/test/3x/torch/algorithms/fp8_quant/unit_tests/test_layers/test_linear.py +++ b/test/3x/torch/algorithms/fp8_quant/unit_tests/test_layers/test_linear.py @@ -134,6 +134,8 @@ def test_linear_dynamic_quantization( scale_format: ScaleFormat, use_hpu_graphs: bool ): + if not use_hpu_graphs and (hp_dtype == torch.bfloat16) and device_type == GAUDI2: + pytest.xfail("[SW-242200] Temporary skip them since the time usage is more than expected.") check_tests_to_skip(scale_method, scale_format, True) N = 1 D_in = 8