search

tenstorrent / tt-forge-fe

The TT-Forge FE is a graph compiler designed to optimize and transform computational graphs for deep learning models, enhancing their performance and efficiency.
https://docs.tenstorrent.com/tt-forge-fe/
Apache License 2.0
20 stars 3 forks source link

RuntimeError: Page size must be divisible by sizeof(uint32_t) because buffers hold uint32_t values #627

Open ashokkumarkannan1 opened 1 week ago

ashokkumarkannan1 commented 1 week ago

Description

Got Page size must be divisible by sizeof(uint32_t) because buffers hold uint32_t values error while running the compiled framework model in Bfloat16

Reproduce

git checkout akannan/fix-invalid-maxpool-pad
git submodule update --recursive
pytest -svv forge/test/mlir/test_ops.py::test_padding_with_maxpool2d[padding1-shape0]

Observed Behaviour

>       all_outputs = run_binary(self.compiled_binary, int(ProgramId.FORWARD), inputs_and_parameters)
E       RuntimeError: TT_FATAL @ /proj_sw/user_dev/akannan/tt-forge-fe/third_party/tt-mlir/third_party/tt-metal/src/tt-metal/tt_metal/impl/buffers/buffer.cpp:41: page_size % sizeof(uint32_t) == 0
E       info:
E       Page size must be divisible by sizeof(uint32_t) because buffers hold uint32_t values
E       backtrace:
E        --- /proj_sw/user_dev/akannan/tt-forge-fe/third_party/tt-mlir/build/lib/SharedLib/libTTMLIR.so(+0x69bab8) [0x7fda40ff3ab8]

Logs

test_padding_with_maxpool2d_2.log

nsmithtt commented 6 days ago

Metal issue with same signature: https://github.com/tenstorrent/tt-metal/issues/13598