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

Explicitly Use __ldg on Pointer Derefs #1800

Closed
wants to merge 1 commit into from
Closed

Explicitly Use __ldg on Pointer Derefs #1800

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
10 changes: 10 additions & 0 deletions include/gridtools/fn/cartesian.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,10 @@
#include "./common_interface.hpp"
#include "./executor.hpp"

#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 350
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We only support cuda >= 11.0, therefore anyway no support for < 350

#include "../common/cuda_type_traits.hpp"
#endif

namespace gridtools::fn {
namespace cartesian::dim {
using i = integral_constant<int, 0>;
Expand Down Expand Up @@ -44,6 +48,12 @@ namespace gridtools::fn {

template <class Tag, class Ptr, class Strides>
GT_FUNCTION auto deref(iterator<Tag, Ptr, Strides> const &it) {
#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 350
if constexpr (std::is_pointer_v<decltype(it.m_ptr)> &&
is_texture_type<std::decay_t<std::remove_pointer_t<decltype(it.m_ptr)>>>::value) {
return __ldg(it.m_ptr);
}
#endif
Comment on lines +51 to +56
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

what about introducing a function like
deref_as_const?

return *it.m_ptr;
}

Expand Down
12 changes: 11 additions & 1 deletion include/gridtools/fn/sid_neighbor_table.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,10 @@
#include "../fn/unstructured.hpp"
#include "../sid/concept.hpp"

#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 350
#include "../common/cuda_type_traits.hpp"
#endif

namespace gridtools::fn::sid_neighbor_table {
namespace sid_neighbor_table_impl_ {
template <class IndexDimension,
Expand Down Expand Up @@ -46,7 +50,13 @@ namespace gridtools::fn::sid_neighbor_table {

sid::shift(ptr, sid::get_stride<IndexDimension>(table.strides), index);
for (std::size_t element_idx = 0; element_idx < MaxNumNeighbors; ++element_idx) {
neighbors[element_idx] = *ptr;
#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 350
if constexpr (std::is_pointer_v<decltype(ptr)> &&
is_texture_type<std::decay_t<std::remove_pointer_t<decltype(ptr)>>>::value)
neighbors[element_idx] = __ldg(ptr);
else
#endif
neighbors[element_idx] = *ptr;
sid::shift(ptr, sid::get_stride<NeighborDimension>(table.strides), 1_c);
}
return neighbors;
Expand Down
13 changes: 12 additions & 1 deletion include/gridtools/fn/unstructured.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,10 @@
#include "./executor.hpp"
#include "./neighbor_table.hpp"

#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 350
#include "../common/cuda_type_traits.hpp"
#endif

namespace gridtools::fn {
namespace unstructured::dim {
using horizontal = integral_constant<int, 0>;
Expand Down Expand Up @@ -80,7 +84,14 @@ namespace gridtools::fn {
GT_FUNCTION constexpr auto deref(iterator<Tag, Ptr, Strides, Domain> const &it) {
GT_PROMISE(can_deref(it));
decltype(auto) stride = host_device::at_key<Tag>(sid::get_stride<dim::horizontal>(it.m_strides));
return *sid::shifted(it.m_ptr, stride, it.m_index);
auto ptr = sid::shifted(it.m_ptr, stride, it.m_index);
#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 350
if constexpr (std::is_pointer_v<decltype(ptr)> &&
is_texture_type<std::decay_t<std::remove_pointer_t<decltype(ptr)>>>::value) {
return __ldg(ptr);
}
#endif
return *ptr;
}

template <class Tag, class Ptr, class Strides, class Domain, class Conn, class Offset>
Expand Down
Loading