Merge pull request #26120 from mshabunin:fix-rvv-init-4.x

RISC-V: remove statically initialized global RVV variables (4.x)
This commit is contained in:
Alexander Smorkalov 2024-09-06 08:21:47 +03:00 committed by GitHub
commit 2a8d4c6025
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -52,22 +52,6 @@ using uint = unsigned int;
using uint64 = unsigned long int;
using int64 = long int;
static const int __cv_rvv_e8m1_nlanes = __riscv_vsetvlmax_e8m1();
static const int __cv_rvv_e16m1_nlanes = __riscv_vsetvlmax_e16m1();
static const int __cv_rvv_e32m1_nlanes = __riscv_vsetvlmax_e32m1();
static const int __cv_rvv_e64m1_nlanes = __riscv_vsetvlmax_e64m1();
static const int __cv_rvv_e8m2_nlanes = __riscv_vsetvlmax_e8m2();
static const int __cv_rvv_e16m2_nlanes = __riscv_vsetvlmax_e16m2();
static const int __cv_rvv_e32m2_nlanes = __riscv_vsetvlmax_e32m2();
static const int __cv_rvv_e64m2_nlanes = __riscv_vsetvlmax_e64m2();
static const int __cv_rvv_e8m4_nlanes = __riscv_vsetvlmax_e8m4();
static const int __cv_rvv_e16m4_nlanes = __riscv_vsetvlmax_e16m4();
static const int __cv_rvv_e32m4_nlanes = __riscv_vsetvlmax_e32m4();
static const int __cv_rvv_e64m4_nlanes = __riscv_vsetvlmax_e64m4();
static const int __cv_rvv_e8m8_nlanes = __riscv_vsetvlmax_e8m8();
static const int __cv_rvv_e16m8_nlanes = __riscv_vsetvlmax_e16m8();
static const int __cv_rvv_e32m8_nlanes = __riscv_vsetvlmax_e32m8();
static const int __cv_rvv_e64m8_nlanes = __riscv_vsetvlmax_e64m8();
template <class T>
struct VTraits;
@ -76,7 +60,7 @@ struct VTraits;
template <> \
struct VTraits<REG> \
{ \
static inline int vlanes() { return __cv_rvv_##SUF##_nlanes; } \
static inline int vlanes() { return __riscv_vsetvlmax_##SUF(); } \
using lane_type = TYP; \
static const int max_nlanes = CV_RVV_MAX_VLEN/SZ; \
};
@ -1813,8 +1797,8 @@ inline int v_scan_forward(const v_float64& a)
// mask: {0,0,0,1, ...} -> {T,T,T,F, ...}
#define OPENCV_HAL_IMPL_RVV_PACK_TRIPLETS(_Tpvec, v_trunc) \
inline _Tpvec v_pack_triplets(const _Tpvec& vec) { \
size_t vl = __cv_rvv_e8m1_nlanes; \
vuint32m1_t one = __riscv_vmv_v_x_u32m1(1, __cv_rvv_e32m1_nlanes); \
size_t vl = VTraits<v_uint8>::vlanes(); \
vuint32m1_t one = __riscv_vmv_v_x_u32m1(1, VTraits<v_uint32>::vlanes()); \
vuint8m1_t zero = __riscv_vmv_v_x_u8m1(0, vl); \
vuint8m1_t mask = __riscv_vreinterpret_u8m1(one); \
return __riscv_vcompress(vec, __riscv_vmseq(v_trunc(__riscv_vslideup(zero, mask, 3, vl)), 0, vl), VTraits<_Tpvec>::vlanes()); \