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

Jump to bottom

fear: Add vrecpeq_u32 #308

Draft
wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

fear: Add vrecpeq_u32 #308

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
372b68d
wip
howjmay Jan 20, 2024
d253430
wip
howjmay Apr 3, 2024
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: 9 additions & 1 deletion neon2rvv.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5853,7 +5853,15 @@ FORCE_INLINE uint8x16_t vcntq_u8(uint8x16_t a) {

FORCE_INLINE float32x2_t vrecpe_f32(float32x2_t a) { return __riscv_vfrec7_v_f32m1(a, 2); }

// FORCE_INLINE uint32x2_t vrecpe_u32(uint32x2_t a);
FORCE_INLINE uint32x2_t vrecpe_u32(uint32x2_t a) {
const uint32_t sign_bit = 0x80000000;
vbool32_t signbit_mask = __riscv_vmsgeu_vx_u32m1_b32(__a, 0x80000000, 2);
const uint32_t input_lower_bound = 0x80000000, input_upper_bound = UINT32_MAX;
const uint32_t estimate_lower_bound = 0x80000000, estimate_upper_bound = 0xff800000;
const uint32_t input_range = input_upper_bound - input_lower_bound;
const uint32_t estimate_range = estimate_upper_bound - estimate_lower_bound;
vuint32m1_t diff = __riscv_vrsub_vx_u32m1(__a, UINT32_MAX, 2);
}

FORCE_INLINE float32x4_t vrecpeq_f32(float32x4_t a) { return __riscv_vfrec7_v_f32m1(a, 4); }

Expand Down
31 changes: 30 additions & 1 deletion tests/impl.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18881,7 +18881,36 @@ result_t test_vrecpe_f32(const NEON2RVV_TEST_IMPL &impl, uint32_t iter) {
#endif // ENABLE_TEST_ALL
}

result_t test_vrecpe_u32(const NEON2RVV_TEST_IMPL &impl, uint32_t iter) { return TEST_UNIMPL; }
result_t test_vrecpe_u32(const NEON2RVV_TEST_IMPL &impl, uint32_t iter) {
// refer: https://community.arm.com/support-forums/f/armds-forum/930/division-with-neon
// https://www.cnblogs.com/pepetang/p/7777243.html
const uint32_t *_a = (const uint32_t *)impl.test_cases_int_pointer1;
uint32_t _c[2];
// const uint32_t sign_bit = 0x80000000;
// const uint32_t input_lower_bound = 0x80000000, input_upper_bound = UINT32_MAX;
// const uint32_t estimate_lower_bound = 0x80000000, estimate_upper_bound = 0xff800000;
// const uint32_t input_range = input_upper_bound - input_lower_bound;
// const uint32_t estimate_range = estimate_upper_bound - estimate_lower_bound;
// for (int i = 0; i < 2; i++) {
// if (!(_a[i] & sign_bit)) {
// _c[i] = UINT32_MAX;
// } else {
// uint32_t diff = UINT32_MAX - _a[i];
// _c[i] = (diff / (input_range) * (estimate_range) + estimate_lower_bound);
// }
// }
for (int i = 0; i < 2; i++) {
x_normalized = x <<
}


uint32x2_t a = vld1_u32(_a);
uint32x2_t c = vrecpe_u32(a);
// print_u32_64("_a", _a);
// print_u32_64("_c", _c);
// print_u32_64("-c", c);
return validate_uint32(c, _c[0], _c[1]);
}

result_t test_vrecpeq_f32(const NEON2RVV_TEST_IMPL &impl, uint32_t iter) {
#ifdef ENABLE_TEST_ALL
Expand Down
Loading