Skip to content

Fix implicit float-to-double promotion. #15957

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

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
+8 −6
Open

Fix implicit float-to-double promotion. #15957

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
4 changes: 2 additions & 2 deletions kernels/portable/cpu/scalar_utils.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -221,8 +221,8 @@ bool extract_scalar(Scalar scalar, FLOAT_T* out_val) {
// be represented when FLOAT_T == float. float can, however, represent
// infinite and NaN values.
if (std::isfinite(val) &&
(val < std::numeric_limits<FLOAT_T>::lowest() ||
val > std::numeric_limits<FLOAT_T>::max())) {
(val < static_cast<double>(std::numeric_limits<FLOAT_T>::lowest()) ||
val > static_cast<double>(std::numeric_limits<FLOAT_T>::max()))) {
// PyTorch's implementation of clamp() raises an exception if the min/max
// values cannot be represented as the dtype, so we should fail too.
return false;
Expand Down
2 changes: 1 addition & 1 deletion kernels/portable/cpu/util/distance_util.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -116,7 +116,7 @@ void pdist(const Tensor& in, Tensor& out, double p) {
pdist<CTYPE, L1<CTYPE>>(in, out, p);
} else if (p == 2.0) {
pdist<CTYPE, L2<CTYPE>>(in, out, p);
} else if (p == INFINITY) {
} else if (p == static_cast<double>(INFINITY)) {
pdist<CTYPE, Linf<CTYPE>>(in, out, p);
} else {
pdist<CTYPE, Lp<CTYPE>>(in, out, p);
Expand Down
3 changes: 2 additions & 1 deletion kernels/portable/cpu/util/math_util.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,7 +47,8 @@ template <
type = true>
FLOAT_T floor_divide(FLOAT_T a, FLOAT_T b) {
if (b == 0) {
return std::signbit(a) ? -INFINITY : INFINITY;
return std::signbit(a) ? static_cast<FLOAT_T>(-INFINITY)
: static_cast<FLOAT_T>(INFINITY);
}
const auto mod = std::fmod(a, b);
auto div = (a - mod) / b;
Expand Down
5 changes: 3 additions & 2 deletions runtime/core/exec_aten/testing_util/tensor_util.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,8 +54,9 @@ bool element_is_close(const T a, const T b, double rtol, double atol) {
return false;
}
} else {
auto allowed_error = atol + std::abs(rtol * b);
auto actual_error = std::abs(a - b);
const double allowed_error =
atol + std::abs(rtol * static_cast<double>(b));
const double actual_error = static_cast<double>(std::abs(a - b));
if (!std::isfinite(actual_error) || actual_error > allowed_error) {
return false;
}
Expand Down