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

[BACKPORT]: Provide backfills for missing __half functionality #1544

Merged
merged 1 commit into from
Mar 19, 2024
Merged

[BACKPORT]: Provide backfills for missing __half functionality #1544

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
2 changes: 1 addition & 1 deletion libcudacxx/include/cuda/std/detail/libcxx/include/__cuda/cmath_nvbf16.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,7 +50,7 @@ inline _LIBCUDACXX_INLINE_VISIBILITY __nv_bfloat16 sinh(__nv_bfloat16 __v)

inline _LIBCUDACXX_INLINE_VISIBILITY __nv_bfloat16 cos(__nv_bfloat16 __v)
{
NV_IF_ELSE_TARGET(NV_IS_DEVICE, (return hcos(__v);), (return __nv_bfloat16(::cos(float(__v)));))
NV_IF_ELSE_TARGET(NV_IS_DEVICE, (return ::hcos(__v);), (return __nv_bfloat16(::cos(float(__v)));))
}

inline _LIBCUDACXX_INLINE_VISIBILITY __nv_bfloat16 cosh(__nv_bfloat16 __v)
Expand Down
57 changes: 29 additions & 28 deletions libcudacxx/include/cuda/std/detail/libcxx/include/__cuda/cmath_nvfp16.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,29 +36,33 @@ _LIBCUDACXX_BEGIN_NAMESPACE_STD
// trigonometric functions
inline _LIBCUDACXX_INLINE_VISIBILITY __half sin(__half __v)
{
NV_IF_ELSE_TARGET(NV_IS_DEVICE, (return hsin(__v);), ({
float __vf = __v;
__vf = ::sin(__vf);
__half_raw __ret_repr = ::__float2half_rn(__vf);
NV_IF_ELSE_TARGET(NV_PROVIDES_SM_53, (
return ::hsin(__v);
), (
{
float __vf = __v;
__vf = ::sin(__vf);
__half_raw __ret_repr = ::__float2half_rn(__vf);

uint16_t __repr = __half_raw(__v).x;
switch (__repr)
{
case 12979:
case 45747:
__ret_repr.x += 1;
break;
uint16_t __repr = __half_raw(__v).x;
switch (__repr)
{
case 12979:
case 45747:
__ret_repr.x += 1;
break;

case 23728:
case 56496:
__ret_repr.x -= 1;
break;
case 23728:
case 56496:
__ret_repr.x -= 1;
break;

default:;
}
default:;
}

return __ret_repr;
}))
return __ret_repr;
}
))
}

inline _LIBCUDACXX_INLINE_VISIBILITY __half sinh(__half __v)
Expand All @@ -69,10 +73,9 @@ inline _LIBCUDACXX_INLINE_VISIBILITY __half sinh(__half __v)
// clang-format off
inline _LIBCUDACXX_INLINE_VISIBILITY __half cos(__half __v)
{
NV_IF_ELSE_TARGET(NV_IS_DEVICE,
(
NV_IF_ELSE_TARGET(NV_PROVIDES_SM_53, (
return ::hcos(__v);
),(
), (
{
float __vf = __v;
__vf = ::cos(__vf);
Expand Down Expand Up @@ -103,10 +106,9 @@ inline _LIBCUDACXX_INLINE_VISIBILITY __half cosh(__half __v)
// clang-format off
inline _LIBCUDACXX_INLINE_VISIBILITY __half exp(__half __v)
{
NV_IF_ELSE_TARGET(NV_IS_DEVICE,
(
NV_IF_ELSE_TARGET(NV_PROVIDES_SM_53, (
return ::hexp(__v);
),(
), (
{
float __vf = __v;
__vf = ::exp(__vf);
Expand Down Expand Up @@ -142,10 +144,9 @@ inline _LIBCUDACXX_INLINE_VISIBILITY __half atan2(__half __x, __half __y)
// clang-format off
inline _LIBCUDACXX_INLINE_VISIBILITY __half log(__half __x)
{
NV_IF_ELSE_TARGET(NV_IS_DEVICE,
(
NV_IF_ELSE_TARGET(NV_PROVIDES_SM_53, (
return ::hlog(__x);
),(
), (
{
float __vf = __x;
__vf = ::log(__vf);
Expand Down
Loading