Skip to content

Refactor extraction of tuning policy selectors #8975

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
bernhardmgruber wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Refactor extraction of tuning policy selectors #8975

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
38 changes: 14 additions & 24 deletions cub/cub/device/device_adjacent_difference.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -635,11 +635,9 @@ struct DeviceAdjacentDifference
{
_CCCL_NVTX_RANGE_SCOPE("cub::DeviceAdjacentDifference::SubtractLeftCopy");

using OffsetT = detail::choose_offset_t<NumItemsT>;
using default_policy_selector = detail::adjacent_difference::policy_selector_from_types<InputIteratorT, false>;

return detail::dispatch_with_env_and_tuning<default_policy_selector>(
env, [&](auto policy_selector, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
using OffsetT = detail::choose_offset_t<NumItemsT>;
return detail::dispatch_with_env(
env, [&](auto tuning_env, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
return detail::adjacent_difference::dispatch<MayAlias::No, ReadOption::Left>(
d_temp_storage,
temp_storage_bytes,
Expand All @@ -648,7 +646,7 @@ struct DeviceAdjacentDifference
static_cast<OffsetT>(num_items),
difference_op,
stream,
policy_selector);
tuning_env);
});
}

Expand Down Expand Up @@ -721,11 +719,8 @@ struct DeviceAdjacentDifference
_CCCL_NVTX_RANGE_SCOPE("cub::DeviceAdjacentDifference::SubtractLeft");

using OffsetT = detail::choose_offset_t<NumItemsT>;
using default_policy_selector =
detail::adjacent_difference::policy_selector_from_types<RandomAccessIteratorT, true>;

return detail::dispatch_with_env_and_tuning<default_policy_selector>(
env, [&](auto policy_selector, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
return detail::dispatch_with_env(
env, [&](auto tuning_env, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
return detail::adjacent_difference::dispatch<MayAlias::Yes, ReadOption::Left>(
d_temp_storage,
temp_storage_bytes,
Expand All @@ -734,7 +729,7 @@ struct DeviceAdjacentDifference
static_cast<OffsetT>(num_items),
difference_op,
stream,
policy_selector);
tuning_env);
});
}

Expand Down Expand Up @@ -817,11 +812,9 @@ struct DeviceAdjacentDifference
{
_CCCL_NVTX_RANGE_SCOPE("cub::DeviceAdjacentDifference::SubtractRightCopy");

using OffsetT = detail::choose_offset_t<NumItemsT>;
using default_policy_selector = detail::adjacent_difference::policy_selector_from_types<InputIteratorT, false>;

return detail::dispatch_with_env_and_tuning<default_policy_selector>(
env, [&](auto policy_selector, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
using OffsetT = detail::choose_offset_t<NumItemsT>;
return detail::dispatch_with_env(
env, [&](auto tuning_env, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
return detail::adjacent_difference::dispatch<MayAlias::No, ReadOption::Right>(
d_temp_storage,
temp_storage_bytes,
Expand All @@ -830,7 +823,7 @@ struct DeviceAdjacentDifference
static_cast<OffsetT>(num_items),
difference_op,
stream,
policy_selector);
tuning_env);
});
}

Expand Down Expand Up @@ -903,11 +896,8 @@ struct DeviceAdjacentDifference
_CCCL_NVTX_RANGE_SCOPE("cub::DeviceAdjacentDifference::SubtractRight");

using OffsetT = detail::choose_offset_t<NumItemsT>;
using default_policy_selector =
detail::adjacent_difference::policy_selector_from_types<RandomAccessIteratorT, true>;

return detail::dispatch_with_env_and_tuning<default_policy_selector>(
env, [&](auto policy_selector, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
return detail::dispatch_with_env(
env, [&](auto tuning_env, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
return detail::adjacent_difference::dispatch<MayAlias::Yes, ReadOption::Right>(
d_temp_storage,
temp_storage_bytes,
Expand All @@ -916,7 +906,7 @@ struct DeviceAdjacentDifference
static_cast<OffsetT>(num_items),
difference_op,
stream,
policy_selector);
tuning_env);
});
}
};
Expand Down
17 changes: 7 additions & 10 deletions cub/cub/device/device_merge.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -192,10 +192,8 @@ struct DeviceMerge
{
_CCCL_NVTX_RANGE_SCOPE("cub::DeviceMerge::MergeKeys");

using default_policy_selector =
detail::merge::policy_selector_from_types<detail::it_value_t<KeyIteratorIn1>, NullType, int64_t>;
return detail::dispatch_with_env_and_tuning<default_policy_selector>(
env, [&](auto policy_selector, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
return detail::dispatch_with_env(
env, [&](auto tuning_env, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
return detail::merge::dispatch(
d_temp_storage,
temp_storage_bytes,
Expand All @@ -209,7 +207,7 @@ struct DeviceMerge
static_cast<NullType*>(nullptr),
compare_op,
stream,
policy_selector);
tuning_env);
});
}

Expand Down Expand Up @@ -416,10 +414,9 @@ struct DeviceMerge
EnvT env = {})
{
_CCCL_NVTX_RANGE_SCOPE("cub::DeviceMerge::MergePairs");
using default_policy_selector = detail::merge::
policy_selector_from_types<detail::it_value_t<KeyIteratorIn1>, detail::it_value_t<ValueIteratorIn1>, int64_t>;
return detail::dispatch_with_env_and_tuning<default_policy_selector>(
env, [&](auto policy_selector, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {

return detail::dispatch_with_env(
env, [&](auto tuning_env, void* d_temp_storage, size_t& temp_storage_bytes, cudaStream_t stream) {
return detail::merge::dispatch(
d_temp_storage,
temp_storage_bytes,
Expand All @@ -433,7 +430,7 @@ struct DeviceMerge
values_out,
compare_op,
stream,
policy_selector);
tuning_env);
});
}
};
Expand Down
Loading
Loading