Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Rewrite Universal Intrinsic code by using new API: ImgProc module Part 2 #24132

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
merged 5 commits into from
Sep 19, 2023
Merged

Rewrite Universal Intrinsic code by using new API: ImgProc module Part 2 #24132

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
83341d4
Rewrite Universal Intrinsic code by using new API: ImgProc module par…
hanliutong Aug 9, 2023
ee5c1f5
Fix compile errors for GCC 13.
hanliutong Aug 9, 2023
ae416d6
Use lambda instead of loop unrolling.
hanliutong Aug 29, 2023
a7e1deb
Revert unnecessary modification.
hanliutong Aug 29, 2023
4c2eded
Fix color_lab.cpp.
hanliutong Sep 16, 2023
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
67 changes: 66 additions & 1 deletion modules/core/include/opencv2/core/hal/intrin.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -745,7 +745,22 @@ namespace CV__SIMD_NAMESPACE {
inline _Tpvec v_add(const _Tpvec& f1, const _Tpvec& f2, const Args&... vf) { \
return v_add(f1 + f2, vf...); \
}
#define OPENCV_HAL_WRAP_SHIFT_OP(_Tpvec) \
inline _Tpvec v_shr(const _Tpvec& a, int n) \
{ \
return a >> n; \
} \
inline _Tpvec v_shl(const _Tpvec& a, int n) \
{ \
return a << n; \
}

OPENCV_HAL_WRAP_SHIFT_OP(v_uint16)
OPENCV_HAL_WRAP_SHIFT_OP(v_uint32)
OPENCV_HAL_WRAP_SHIFT_OP(v_uint64)
OPENCV_HAL_WRAP_SHIFT_OP(v_int16)
OPENCV_HAL_WRAP_SHIFT_OP(v_int32)
OPENCV_HAL_WRAP_SHIFT_OP(v_int64)
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_uint8)
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_uint16)
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_uint32)
Expand All @@ -769,6 +784,12 @@ namespace CV__SIMD_NAMESPACE {
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_int32x4)
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_int64x2)
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_float32x4)
OPENCV_HAL_WRAP_SHIFT_OP(v_uint16x8)
OPENCV_HAL_WRAP_SHIFT_OP(v_uint32x4)
OPENCV_HAL_WRAP_SHIFT_OP(v_uint64x2)
OPENCV_HAL_WRAP_SHIFT_OP(v_int16x8)
OPENCV_HAL_WRAP_SHIFT_OP(v_int32x4)
OPENCV_HAL_WRAP_SHIFT_OP(v_int64x2)
#if CV_SIMD_64F
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_float64x2)
#endif
Expand All @@ -784,6 +805,12 @@ namespace CV__SIMD_NAMESPACE {
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_int32x8)
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_int64x4)
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_float32x8)
OPENCV_HAL_WRAP_SHIFT_OP(v_uint16x16)
OPENCV_HAL_WRAP_SHIFT_OP(v_uint32x8)
OPENCV_HAL_WRAP_SHIFT_OP(v_uint64x4)
OPENCV_HAL_WRAP_SHIFT_OP(v_int16x16)
OPENCV_HAL_WRAP_SHIFT_OP(v_int32x8)
OPENCV_HAL_WRAP_SHIFT_OP(v_int64x4)
#if CV_SIMD_64F
OPENCV_HAL_WRAP_BIN_OP_ADDSUB(v_float64x4)
#endif
Expand All @@ -801,7 +828,9 @@ namespace CV__SIMD_NAMESPACE {
inline _Tpvec v_xor(const _Tpvec& a, const _Tpvec& b) \
{ \
return a ^ b; \
} \
}

#define OPENCV_HAL_WRAP_NOT_OP(_Tpvec) \
inline _Tpvec v_not(const _Tpvec& a) \
{ \
return ~a; \
Expand All @@ -815,6 +844,18 @@ namespace CV__SIMD_NAMESPACE {
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_int16)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_int32)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_int64)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_float32)
OPENCV_HAL_WRAP_NOT_OP(v_uint8)
OPENCV_HAL_WRAP_NOT_OP(v_uint16)
OPENCV_HAL_WRAP_NOT_OP(v_uint32)
OPENCV_HAL_WRAP_NOT_OP(v_uint64)
OPENCV_HAL_WRAP_NOT_OP(v_int8)
OPENCV_HAL_WRAP_NOT_OP(v_int16)
OPENCV_HAL_WRAP_NOT_OP(v_int32)
OPENCV_HAL_WRAP_NOT_OP(v_int64)
#if CV_SIMD_64F
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_float64)
#endif
#if CV_SIMD_WIDTH != 16/*128*/ && CV_SIMD128
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_uint8x16)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_uint16x8)
Expand All @@ -824,6 +865,18 @@ namespace CV__SIMD_NAMESPACE {
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_int16x8)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_int32x4)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_int64x2)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_float32x4)
OPENCV_HAL_WRAP_NOT_OP(v_uint8x16)
OPENCV_HAL_WRAP_NOT_OP(v_uint16x8)
OPENCV_HAL_WRAP_NOT_OP(v_uint32x4)
OPENCV_HAL_WRAP_NOT_OP(v_uint64x2)
OPENCV_HAL_WRAP_NOT_OP(v_int8x16)
OPENCV_HAL_WRAP_NOT_OP(v_int16x8)
OPENCV_HAL_WRAP_NOT_OP(v_int32x4)
OPENCV_HAL_WRAP_NOT_OP(v_int64x2)
#if CV_SIMD_64F
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_float64x2)
#endif
#endif
#if CV_SIMD_WIDTH != 32/*256*/ && CV_SIMD256
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_uint8x32)
Expand All @@ -834,6 +887,18 @@ namespace CV__SIMD_NAMESPACE {
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_int16x16)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_int32x8)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_int64x4)
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_float32x8)
OPENCV_HAL_WRAP_NOT_OP(v_uint8x32)
OPENCV_HAL_WRAP_NOT_OP(v_uint16x16)
OPENCV_HAL_WRAP_NOT_OP(v_uint32x8)
OPENCV_HAL_WRAP_NOT_OP(v_uint64x4)
OPENCV_HAL_WRAP_NOT_OP(v_int8x32)
OPENCV_HAL_WRAP_NOT_OP(v_int16x16)
OPENCV_HAL_WRAP_NOT_OP(v_int32x8)
OPENCV_HAL_WRAP_NOT_OP(v_int64x4)
#if CV_SIMD_64F
OPENCV_HAL_WRAP_BIN_OP_LOGIC(v_float64x4)
#endif
#endif

#define OPENCV_HAL_WRAP_BIN_OP_MUL(_Tpvec) \
Expand Down
1 change: 1 addition & 0 deletions modules/core/include/opencv2/core/hal/intrin_rvv_compat_overloaded.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,6 +45,7 @@ OPENCV_HAL_IMPL_RVV_FUN_LOXEI(vuint8m2_t, u8m2, vuint8m2_t, i8)
OPENCV_HAL_IMPL_RVV_FUN_LOXEI(vuint8m4_t, u8m4, vuint8m4_t, i8)
OPENCV_HAL_IMPL_RVV_FUN_LOXEI(vuint8m8_t, u8m8, vuint8m8_t, i8)
OPENCV_HAL_IMPL_RVV_FUN_LOXEI(vfloat32m1_t, f32m1, vuint32m1_t, i32)
OPENCV_HAL_IMPL_RVV_FUN_LOXEI(vuint32m1_t, u32m1, vuint32m1_t, i32)
#if CV_SIMD_SCALABLE_64F
OPENCV_HAL_IMPL_RVV_FUN_LOXEI(vfloat64m1_t, f64m1, vuint32mf2_t, i32)
#endif
Expand Down
31 changes: 27 additions & 4 deletions modules/core/include/opencv2/core/hal/intrin_rvv_scalable.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -475,6 +475,25 @@ OPENCV_HAL_IMPL_RVV_LUT(v_float32, float, m1)
OPENCV_HAL_IMPL_RVV_LUT(v_float64, double, mf2)
#endif

#define OPENCV_HAL_IMPL_RVV_LUT_VEC(_Tpvec, _Tp) \
inline _Tpvec v_lut(const _Tp* tab, const v_int32& vidx) \
{ \
v_uint32 vidx_ = vmul(vreinterpret_u32m1(vidx), sizeof(_Tp), VTraits<v_int32>::vlanes()); \
return vloxei32(tab, vidx_, VTraits<_Tpvec>::vlanes()); \
}
OPENCV_HAL_IMPL_RVV_LUT_VEC(v_float32, float)
OPENCV_HAL_IMPL_RVV_LUT_VEC(v_int32, int)
OPENCV_HAL_IMPL_RVV_LUT_VEC(v_uint32, unsigned)

#if CV_SIMD_SCALABLE_64F
inline v_float64 v_lut(const double* tab, const v_int32& vidx) \
{ \
vuint32mf2_t vidx_ = vmul(vlmul_trunc_u32mf2(vreinterpret_u32m1(vidx)), sizeof(double), VTraits<v_float64>::vlanes()); \
return vloxei32(tab, vidx_, VTraits<v_float64>::vlanes()); \
}
#endif


inline v_uint8 v_lut(const uchar* tab, const int* idx) { return v_reinterpret_as_u8(v_lut((schar*)tab, idx)); }
inline v_uint8 v_lut_pairs(const uchar* tab, const int* idx) { return v_reinterpret_as_u8(v_lut_pairs((schar*)tab, idx)); }
inline v_uint8 v_lut_quads(const uchar* tab, const int* idx) { return v_reinterpret_as_u8(v_lut_quads((schar*)tab, idx)); }
Expand Down Expand Up @@ -690,23 +709,27 @@ inline v_float64 v_not (const v_float64& a) \


////////////// Bitwise shifts //////////////
/* Usage
1. v_shl<N>(vec);
2. v_shl(vec, N); // instead of vec << N, when N is non-constant.
*/

#define OPENCV_HAL_IMPL_RVV_UNSIGNED_SHIFT_OP(_Tpvec, vl) \
template<int n> inline _Tpvec v_shl(const _Tpvec& a) \
template<int s = 0> inline _Tpvec v_shl(const _Tpvec& a, int n = s) \
{ \
return _Tpvec(vsll(a, uint8_t(n), vl)); \
} \
template<int n> inline _Tpvec v_shr(const _Tpvec& a) \
template<int s = 0> inline _Tpvec v_shr(const _Tpvec& a, int n = s) \
{ \
return _Tpvec(vsrl(a, uint8_t(n), vl)); \
}

#define OPENCV_HAL_IMPL_RVV_SIGNED_SHIFT_OP(_Tpvec, vl) \
template<int n> inline _Tpvec v_shl(const _Tpvec& a) \
template<int s = 0> inline _Tpvec v_shl(const _Tpvec& a, int n = s) \
{ \
return _Tpvec(vsll(a, uint8_t(n), vl)); \
} \
template<int n> inline _Tpvec v_shr(const _Tpvec& a) \
template<int s = 0> inline _Tpvec v_shr(const _Tpvec& a, int n = s) \
{ \
return _Tpvec(vsra(a, uint8_t(n), vl)); \
}
Expand Down
Loading