You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by GitBox <gi...@apache.org> on 2022/10/19 19:54:04 UTC
[GitHub] [tvm] supersat commented on pull request #13002: [Hexagon] Fix chunk address table generation in chunkify_hwio_16b
supersat commented on PR #13002:
URL: https://github.com/apache/tvm/pull/13002#issuecomment-1284501589
All tests pass on hardware after zeroing out memory before returning from `VTCMAllocation::VTCMAllocation()`:
```=============================================================================== test session starts ================================================================================
platform linux -- Python 3.8.10, pytest-6.2.5, py-1.11.0, pluggy-1.0.0 -- /usr/bin/python3
cachedir: .pytest_cache
rootdir: /home/kkoscher/Projects/tvm
plugins: xdist-2.5.0, forked-1.4.0
collected 19 items
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x8x4x3-hwio3x3x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x10x14x3-hwio3x3x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x14x6x3-hwio3x3x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x14x6x3-hwio3x3x3x64-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x14x6x3-hwio5x5x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x8x8x3-hwio2x2x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x14x6x64-hwio3x3x64x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x4x4x40-hwio3x3x40x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x4x4x3-hwio3x3x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x5x5x3-hwio3x3x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x6x6x3-hwio3x3x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x7x7x3-hwio3x3x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x8x8x3-hwio3x3x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x8x8x3-hwio5x5x3x3-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x8x8x64-hwio2x2x64x64-stride1x1] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x8x4x3-hwio3x3x3x3-stride2x2] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x14x6x3-hwio3x3x3x64-stride2x2] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x14x6x3-hwio5x5x3x3-stride2x2] PASSED
tests/python/contrib/test_hexagon/topi/test_conv2d_fp16_intrin.py::TestConv2dIntrin::test_conv2d[offset0x0-nhwc1x8x8x3-hwio2x2x3x3-stride2x2] PASSED
=============================================================================== 19 passed in 40.70s ================================================================================```
Properly zeroing memory should likely be done in the conv2d kernel rather than relying on VTCMAllocator to do that every time, but that can be addressed in another PR, so this one LGTM.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org