Skip to content

Implement batch_bool::mask() for riscv #1238

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
serge-sans-paille merged 2 commits into from
Jan 5, 2026
Merged

Implement batch_bool::mask() for riscv #1238

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
af74418
[riscv] Remove warning about shadowing ctx
serge-sans-paille Jan 4, 2026
2b4f903
Implement batch_bool::mask() for riscv
serge-sans-paille Dec 30, 2025
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
63 changes: 57 additions & 6 deletions include/xsimd/arch/xsimd_rvv.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,7 @@
#include <type_traits>
#include <utility>

#include "../types/xsimd_batch_constant.hpp"
#include "../types/xsimd_rvv_register.hpp"
#include "xsimd_constants.hpp"

Expand Down Expand Up @@ -135,10 +136,9 @@
template <class Ret, class... Args> \
struct impl<KEY, Ret(Args...)> \
{ \
using ctx = ctx<KEY>; \
constexpr Ret operator()(Args... args) const noexcept \
{ \
return CALLEE(args..., ctx::vl); \
return CALLEE(args..., ctx<KEY>::vl); \
}; \
};
#define XSIMD_RVV_WRAPPER_NOVL(KEY, CALLEE, ...) \
Expand All @@ -154,20 +154,18 @@
template <class Ret, class First, class... Args> \
struct impl<KEY, Ret(First, Args...)> \
{ \
using ctx = ctx<KEY>; \
constexpr Ret operator()(First, Args... args) const noexcept \
{ \
return CALLEE(args..., ctx::vl); \
return CALLEE(args..., ctx<KEY>::vl); \
}; \
};
#define XSIMD_RVV_WRAPPER_DROP_1ST_CUSTOM_ARGS(KEY, CALLEE, SIGNATURE, ...) \
template <class Ret, class First, class... Args> \
struct impl<KEY, Ret(First, Args...)> \
{ \
using ctx = ctx<KEY>; \
constexpr Ret operator()(First, Args... args) const noexcept \
{ \
return CALLEE(__VA_ARGS__, ctx::vl); \
return CALLEE(__VA_ARGS__, ctx<KEY>::vl); \
}; \
};
#define XSIMD_RVV_WRAPPER_DROP_1ST_CUSTOM_ARGS_NOVL(KEY, CALLEE, SIGNATURE, ...) \
Expand Down Expand Up @@ -1507,6 +1505,59 @@ namespace xsimd
const auto mask = abs(arg) < constants::maxflint<batch<T, A>>();
return select(mask, to_float(detail::rvvfcvt_default(arg)), arg, rvv {});
}

// mask
template <class A, class T>
XSIMD_INLINE uint64_t mask(batch_bool<T, A> const& self, requires_arch<common>) noexcept;

template <class A, class T>
XSIMD_INLINE uint64_t mask(batch_bool<T, A> const& self, requires_arch<rvv>) noexcept
{
XSIMD_IF_CONSTEXPR((8 * sizeof(T)) >= batch_bool<T, A>::size)
{
// (A) Easy case: the number of slots fits in T.
const auto zero = detail::broadcast<as_unsigned_integer_t<T>, types::detail::rvv_width_m1>(T(0));
auto ones = detail::broadcast<as_unsigned_integer_t<T>, A::width>(1);
auto iota = detail::rvvid(as_unsigned_integer_t<T> {});
auto upowers = detail::rvvsll(ones, iota);
auto r = __riscv_vredor(self.data.as_mask(), upowers, (typename decltype(zero)::register_type)zero, batch_bool<T, A>::size);
return detail::reduce_scalar<A, as_unsigned_integer_t<T>>(r);
}
else XSIMD_IF_CONSTEXPR((2 * 8 * sizeof(T)) == batch_bool<T, A>::size)
{
// (B) We need two rounds, one for the low part, one for the high part.

struct LowerHalf
{
static constexpr bool get(unsigned i, unsigned n) { return i < n / 2; }
};
struct UpperHalf
{
static constexpr bool get(unsigned i, unsigned n) { return i >= n / 2; }
};

// The low part is similar to the approach in (A).
const auto zero = detail::broadcast<as_unsigned_integer_t<T>, types::detail::rvv_width_m1>(T(0));
auto ones = detail::broadcast<as_unsigned_integer_t<T>, A::width>(1);
auto iota = detail::rvvid(as_unsigned_integer_t<T> {});
auto upowers = detail::rvvsll(ones, iota);
auto low_mask = self & make_batch_bool_constant<T, LowerHalf, A>();
auto r_low = __riscv_vredor(low_mask.data.as_mask(), upowers, (typename decltype(zero)::register_type)zero, batch_bool<T, A>::size);

// The high part requires to slide the upower filter to match the high mask.
upowers = detail::rvvslideup(upowers, upowers, 8 * sizeof(T));
auto high_mask = self & make_batch_bool_constant<T, UpperHalf, A>();
auto r_high = __riscv_vredor(high_mask.data.as_mask(), upowers, (typename decltype(zero)::register_type)zero, batch_bool<T, A>::size);

// Concatenate the two parts.
return (uint64_t)detail::reduce_scalar<A, as_unsigned_integer_t<T>>(r_low) | ((uint64_t)detail::reduce_scalar<A, as_unsigned_integer_t<T>>(r_high) << (8 * sizeof(T)));
}
else
{
// (C) we could generalize (B) but we already cover a lot of case now.
return mask(self, common {});
}
}
} // namespace kernel
} // namespace xsimd

Expand Down
1 change: 1 addition & 0 deletions include/xsimd/types/xsimd_rvv_register.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -391,6 +391,7 @@ namespace xsimd
{
}
operator type() const noexcept { return bool_info::bitcast(value); }
type as_mask() const noexcept { return (type) * this; }
};

template <class T, size_t Width = XSIMD_RVV_BITS>
Expand Down