Skip to content

[PHI]Fix nanmedian #74263

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

Merged
merged 2 commits into from
Jul 30, 2025
Merged

[PHI]Fix nanmedian #74263

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1449d43
fix
fangfangssj Jul 27, 2025
bc8faa9
add test
fangfangssj Jul 27, 2025
File filter

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions paddle/phi/kernels/gpu/nanmedian_kernel.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -258,12 +258,12 @@ void ProcessMedianKernel(const Context& dev_ctx,
int64_t sum =
thrust::reduce(exec_policy, nan_counts_ptr, nan_counts_ptr + pre_dim);
nan_stat_cpu_ptr[0] = sum;
auto maxx_ptr = thrust::max_element(
auto min_nan_ptr = thrust::min_element(
exec_policy, nan_counts_ptr, nan_counts_ptr + pre_dim);
memory_utils::Copy(phi::CPUPlace(),
nan_stat_cpu_ptr + 1,
dev_ctx.GetPlace(),
maxx_ptr,
min_nan_ptr,
sizeof(int64_t),
stream);
// all elements are nan values
Expand All @@ -278,7 +278,7 @@ void ProcessMedianKernel(const Context& dev_ctx,
}

ignore_nan = nan_stat_cpu_ptr[0] > 0;
max_valid_num = nan_stat_cpu_ptr[1];
max_valid_num = stride - nan_stat_cpu_ptr[1];
}

int64_t sort_k = ignore_nan ? max_valid_num : ((stride >> 1) + 1);
Expand Down
15 changes: 15 additions & 0 deletions test/legacy_test/test_nanmedian.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -639,6 +639,21 @@ def init_data(self):
self.axis = -1


class TestNanmedianNan(unittest.TestCase):
def init_data(self):
x_np = np.array([1.4907, np.nan, 1.0593, 1.5696])
self.x = paddle.to_tensor(x_np)
self.expect_out = np.nanmedian(x_np)
self.axis = None

def test_nan(self):
self.init_data()
out = paddle.nanmedian(self.x, axis=self.axis)
np.testing.assert_allclose(
out.numpy(), self.expect_out, rtol=1e-05, equal_nan=True
)


@unittest.skipIf(
not core.is_compiled_with_cuda()
or not core.is_bfloat16_supported(core.CUDAPlace(0)),
Expand Down
Loading