Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Manually add libtorch op test #10758

Open
wants to merge 14 commits into
base: main
Choose a base branch
from
Open

Manually add libtorch op test #10758

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
14 commits
Select commit Hold shift + click to select a range
97653cb
Manually add libtorch op test
t-vi Mar 24, 2022
eb6ad91
fix black difference
t-vi Mar 24, 2022
85aba6c
add USE_LIBTORCH to GPU build config
t-vi Mar 24, 2022
244609b
Update tests/scripts/task_config_build_gpu.sh
t-vi Mar 25, 2022
5efcc99
include dlpack at very top
t-vi Mar 25, 2022
03c8f0c
Merge branch 'main' into libtorch_op_test
t-vi Apr 1, 2022
355c5a4
undo messing with the includes
t-vi Apr 1, 2022
70a8929
reorder includes to avoid redifinition warning
t-vi Apr 1, 2022
e2b70b5
Merge branch 'main' into libtorch_op_test
t-vi Apr 20, 2022
daed2cc
experimentally use self-built cmake like arm for ci-gpu
t-vi Apr 22, 2022
a7e954d
Merge branch 'main' into libtorch_op_test
t-vi May 1, 2022
1b682a5
Merge branch 'main' into libtorch_op_test
t-vi May 20, 2022
9873eaa
use libtorch
t-vi May 20, 2022
b5b83f9
Merge branch 'main' into libtorch_op_test
t-vi May 22, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 4 additions & 5 deletions src/runtime/contrib/libtorch/libtorch_runtime.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,16 +24,15 @@

// we do not want clang to reorder our includes
// clang-format off
#include <tvm/runtime/module.h>
#include <tvm/runtime/ndarray.h>
#include <tvm/runtime/registry.h>
#include <tvm/runtime/contrib/libtorch_runtime.h>

#include <ATen/dlpack.h>
#include <ATen/DLConvertor.h>
#include <torch/csrc/jit/serialization/import.h>
#include <torch/torch.h>

#include <tvm/runtime/module.h>
#include <tvm/runtime/ndarray.h>
#include <tvm/runtime/registry.h>
#include <tvm/runtime/contrib/libtorch_runtime.h>
// clang-format on

#include <cstddef>
Expand Down
3 changes: 3 additions & 0 deletions tests/python/contrib/test_libtorch_ops.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,8 @@
# specific language governing permissions and limitations
# under the License.

import tvm.testing

import pytest

import tvm.relay
Expand All @@ -29,6 +31,7 @@
import_torch_error = str(e)


@tvm.testing.requires_gpu
@pytest.mark.skipif(torch is None, reason=f"PyTorch is not available: {import_torch_error}")
def test_backend():
@torch.jit.script
Expand Down
1 change: 1 addition & 0 deletions tests/scripts/task_config_build_gpu.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,3 +48,4 @@ echo set\(USE_LIBBACKTRACE AUTO\) >> config.cmake
echo set\(USE_CCACHE OFF\) >> config.cmake
echo set\(SUMMARIZE ON\) >> config.cmake
echo set\(HIDE_PRIVATE_SYMBOLS ON\) >> config.cmake
echo set\(USE_LIBTORCH /usr/local/libtorch\) >> config.cmake
3 changes: 3 additions & 0 deletions tests/scripts/task_python_integration.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -64,6 +64,9 @@ run_pytest ctypes ${TVM_INTEGRATION_TESTSUITE_NAME}-integration tests/python/int
run_pytest ctypes ${TVM_INTEGRATION_TESTSUITE_NAME}-contrib tests/python/contrib --ignore=tests/python/contrib/test_ethosu
run_pytest ctypes ${TVM_INTEGRATION_TESTSUITE_NAME}-contrib-test_ethosu tests/python/contrib/test_ethosu -n auto

# we want this to run in particular in integration-gpu because there we have PyTorch
run_pytest cython libtorch tests/python/contrib/test_libtorch_ops.py

# forked is needed because the global registry gets contaminated
TVM_TEST_TARGETS="${TVM_RELAY_TEST_TARGETS:-llvm;cuda}" \
run_pytest ctypes ${TVM_INTEGRATION_TESTSUITE_NAME}-relay tests/python/relay
Expand Down