i wonder whether caffe2 support CUDA10.2 and cudnn10.2. i compile pytorch on win10 and installed successfully.
i could run small regression task successfully. however when run the test_spatial_narrow_as_op.py test, it always fails with the log:
"
[E ..\caffe2\core\init_intrinsics_check.cc:43] CPU feature avx is present on your machine, but the Caffe2 binary is not compiled with it. It means you may not get the full speed of your CPU.
[E ..\caffe2\core\init_intrinsics_check.cc:43] CPU feature avx2 is present on your machine, but the Caffe2 binary is not compiled with it. It means you may not get the full speed of your CPU.
[E ..\caffe2\core\init_intrinsics_check.cc:43] CPU feature fma is present on your machine, but the Caffe2 binary is not compiled with it. It means you may not get the full speed of your CPU.
Found Detectron ops lib: C:\Users\szx3gg\Programs\Python\Python38-64\Lib\site-packages\torch/lib\caffe2_detectron_ops_gpu.dll
c.shape: (2, 256, 35, 87)
F.c.shape: (2, 3, 2, 2)
F

FAIL: test_large_forward (main.SpatialNarrowAsOpTest)

Traceback (most recent call last):
File "test_spatial_narrow_as_op.py", line 70, in test_large_forward
self._run_test(A, B)
File "test_spatial_narrow_as_op.py", line 56, in _run_test
np.testing.assert_allclose(C, C_ref, rtol=1e-5, atol=1e-08)
File "C:\Users\szx3gg\Programs\Python\Python38-64\Lib\site-packages\numpy\testing_private\utils.py", line 1532, in assert_allclose
assert_array_compare(compare, actual, desired, err_msg=str(err_msg),
File "C:\Users\szx3gg\Programs\Python\Python38-64\Lib\site-packages\numpy\testing_private\utils.py", line 846, in assert_array_compare
raise AssertionError(msg)
AssertionError:
Not equal to tolerance rtol=1e-05, atol=1e-08

Mismatched elements: 1559040 / 1559040 (100%)
Max absolute difference: 5.026066
Max relative difference: 1.
x: array([[[[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],
[0., 0., 0., ..., 0., 0., 0.],...
y: array([[[[ 9.423651e-01, 1.103046e+00, -3.403718e-01, ...,
1.799662e+00, 6.770432e-01, -2.352981e-01],
[ 1.366407e+00, 1.219485e+00, -2.637292e-01, ...,...

======================================================================
FAIL: test_small_forward_and_gradient (main.SpatialNarrowAsOpTest)

Traceback (most recent call last):
File "test_spatial_narrow_as_op.py", line 61, in test_small_forward_and_gradient
self._run_test(A, B, check_grad=True)
File "test_spatial_narrow_as_op.py", line 56, in _run_test
np.testing.assert_allclose(C, C_ref, rtol=1e-5, atol=1e-08)
File "C:\Users\szx3gg\Programs\Python\Python38-64\Lib\site-packages\numpy\testing_private\utils.py", line 1532, in assert_allclose
assert_array_compare(compare, actual, desired, err_msg=str(err_msg),
File "C:\Users\szx3gg\Programs\Python\Python38-64\Lib\site-packages\numpy\testing_private\utils.py", line 846, in assert_array_compare
raise AssertionError(msg)
AssertionError:
Not equal to tolerance rtol=1e-05, atol=1e-08

Mismatched elements: 24 / 24 (100%)
Max absolute difference: 2.239709
Max relative difference: 1.
x: array([[[[0., 0.],
[0., 0.]],
...
y: array([[[[-0.423967, -0.105844],
[-0.583256, 1.152265]],
...

Ran 3 tests in 1.238s

FAILED (failures=2)
"