Skip to content

Commit

Permalink
Merge pull request #39 from howjmay/div
Browse files Browse the repository at this point in the history 
feat: Add _mm_div_*
  • Loading branch information
@howjmay
howjmay authored Jan 6, 2024
2 parents e9a0c94 + bcf6d54 commit 9b6ebcf
Show file tree
Hide file tree
Showing 2 changed files with 93 additions and 75 deletions.
26 changes: 22 additions & 4 deletions sse2rvv.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1370,13 +1370,31 @@ FORCE_INLINE __m128i _mm_cvtepu8_epi64(__m128i a) {

// FORCE_INLINE __int64 _mm_cvttss_si64 (__m128 a) {}

// FORCE_INLINE __m128d _mm_div_pd (__m128d a, __m128d b) {}
FORCE_INLINE __m128d _mm_div_pd(__m128d a, __m128d b) {
vfloat64m1_t _a = vreinterpretq_m128d_f64(a);
vfloat64m1_t _b = vreinterpretq_m128d_f64(b);
return vreinterpretq_f64_m128d(__riscv_vfdiv_vv_f64m1(_a, _b, 2));
}

// FORCE_INLINE __m128 _mm_div_ps (__m128 a, __m128 b) {}
FORCE_INLINE __m128 _mm_div_ps(__m128 a, __m128 b) {
vfloat32m1_t _a = vreinterpretq_m128_f32(a);
vfloat32m1_t _b = vreinterpretq_m128_f32(b);
return vreinterpretq_f32_m128(__riscv_vfdiv_vv_f32m1(_a, _b, 4));
}

// FORCE_INLINE __m128d _mm_div_sd (__m128d a, __m128d b) {}
FORCE_INLINE __m128d _mm_div_sd(__m128d a, __m128d b) {
vfloat64m1_t _a = vreinterpretq_m128d_f64(a);
vfloat64m1_t _b = vreinterpretq_m128d_f64(b);
vfloat64m1_t div = __riscv_vfdiv_vv_f64m1(_a, _b, 2);
return vreinterpretq_f64_m128d(__riscv_vslideup_vx_f64m1(_a, div, 0, 1));
}

// FORCE_INLINE __m128 _mm_div_ss (__m128 a, __m128 b) {}
FORCE_INLINE __m128 _mm_div_ss(__m128 a, __m128 b) {
vfloat32m1_t _a = vreinterpretq_m128_f32(a);
vfloat32m1_t _b = vreinterpretq_m128_f32(b);
vfloat32m1_t div = __riscv_vfdiv_vv_f32m1(_a, _b, 4);
return vreinterpretq_f32_m128(__riscv_vslideup_vx_f32m1(_a, div, 0, 1));
}

// FORCE_INLINE __m128d _mm_dp_pd (__m128d a, __m128d b, const int imm8) {}

Expand Down
142 changes: 71 additions & 71 deletions tests/impl.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1495,7 +1495,7 @@ result_t test_mm_comineq_ss(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
}

result_t test_mm_cvt_pi2ps(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
// #ifdef ENABLE_TEST_ALL
#ifdef ENABLE_TEST_ALL
const float *_a = impl.test_cases_float_pointer1;
const int32_t *_b = impl.test_cases_int_pointer2;

Expand All @@ -1509,13 +1509,13 @@ result_t test_mm_cvt_pi2ps(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
__m128 c = _mm_cvt_pi2ps(a, b);

return validate_float(c, dx, dy, dz, dw);
// #else
#else
return TEST_UNIMPL;
// #endif // ENABLE_TEST_ALL
#endif // ENABLE_TEST_ALL
}

result_t test_mm_cvt_ps2pi(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
// #ifdef ENABLE_TEST_ALL
#ifdef ENABLE_TEST_ALL
const float *_a = impl.test_cases_float_pointer1;
int32_t d[2];

Expand All @@ -1527,13 +1527,13 @@ result_t test_mm_cvt_ps2pi(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
__m64 ret = _mm_cvt_ps2pi(a);

return VALIDATE_INT32_M64(ret, d);
// #else
#else
return TEST_UNIMPL;
// #endif // ENABLE_TEST_ALL
#endif // ENABLE_TEST_ALL
}

result_t test_mm_cvt_si2ss(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
// #ifdef ENABLE_TEST_ALL
#ifdef ENABLE_TEST_ALL
const float *_a = impl.test_cases_float_pointer1;
const int32_t b = *impl.test_cases_int_pointer2;

Expand All @@ -1546,9 +1546,9 @@ result_t test_mm_cvt_si2ss(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
__m128 c = _mm_cvt_si2ss(a, b);

return validate_float(c, dx, dy, dz, dw);
// #else
#else
return TEST_UNIMPL;
// #endif // ENABLE_TEST_ALL
#endif // ENABLE_TEST_ALL
}

result_t test_mm_cvt_ss2si(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
Expand Down Expand Up @@ -1986,42 +1986,42 @@ result_t test_mm_cvttss_si64(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
}

result_t test_mm_div_ps(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
// #ifdef ENABLE_TEST_ALL
// const float *_a = impl.test_cases_float_pointer1;
// const float *_b = impl.test_cases_float_pointer2;
// float f0 = _a[0] / _b[0];
// float f1 = _a[1] / _b[1];
// float f2 = _a[2] / _b[2];
// float f3 = _a[3] / _b[3];
//
// __m128 a = load_m128(_a);
// __m128 b = load_m128(_b);
// __m128 c = _mm_div_ps(a, b);
//
// return validate_float(c, f0, f1, f2, f3);
// #else
#ifdef ENABLE_TEST_ALL
const float *_a = impl.test_cases_float_pointer1;
const float *_b = impl.test_cases_float_pointer2;
float f0 = _a[0] / _b[0];
float f1 = _a[1] / _b[1];
float f2 = _a[2] / _b[2];
float f3 = _a[3] / _b[3];

__m128 a = load_m128(_a);
__m128 b = load_m128(_b);
__m128 c = _mm_div_ps(a, b);

return validate_float(c, f0, f1, f2, f3);
#else
return TEST_UNIMPL;
// #endif // ENABLE_TEST_ALL
#endif // ENABLE_TEST_ALL
}

result_t test_mm_div_ss(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
// #ifdef ENABLE_TEST_ALL
// const float *_a = impl.test_cases_float_pointer1;
// const float *_b = impl.test_cases_float_pointer2;
//
// float d0 = _a[0] / _b[0];
// float d1 = _a[1];
// float d2 = _a[2];
// float d3 = _a[3];
//
// __m128 a = load_m128(_a);
// __m128 b = load_m128(_b);
// __m128 c = _mm_div_ss(a, b);
//
// return validate_float(c, d0, d1, d2, d3);
// #else
#ifdef ENABLE_TEST_ALL
const float *_a = impl.test_cases_float_pointer1;
const float *_b = impl.test_cases_float_pointer2;

float d0 = _a[0] / _b[0];
float d1 = _a[1];
float d2 = _a[2];
float d3 = _a[3];

__m128 a = load_m128(_a);
__m128 b = load_m128(_b);
__m128 c = _mm_div_ss(a, b);

return validate_float(c, d0, d1, d2, d3);
#else
return TEST_UNIMPL;
// #endif // ENABLE_TEST_ALL
#endif // ENABLE_TEST_ALL
}

result_t test_mm_extract_pi16(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
Expand Down Expand Up @@ -5545,42 +5545,42 @@ result_t test_mm_cvttsd_si64x(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
}

result_t test_mm_div_pd(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
// #ifdef ENABLE_TEST_ALL
// const double *_a = (const double *)impl.test_cases_float_pointer1;
// const double *_b = (const double *)impl.test_cases_float_pointer2;
// double d0 = 0.0, d1 = 0.0;
//
// if (_b[0] != 0.0)
// d0 = _a[0] / _b[0];
// if (_b[1] != 0.0)
// d1 = _a[1] / _b[1];
//
// __m128d a = load_m128d(_a);
// __m128d b = load_m128d(_b);
// __m128d c = _mm_div_pd(a, b);
// return validate_double(c, d0, d1);
// #else
#ifdef ENABLE_TEST_ALL
const double *_a = (const double *)impl.test_cases_float_pointer1;
const double *_b = (const double *)impl.test_cases_float_pointer2;
double d0 = 0.0, d1 = 0.0;

if (_b[0] != 0.0)
d0 = _a[0] / _b[0];
if (_b[1] != 0.0)
d1 = _a[1] / _b[1];

__m128d a = load_m128d(_a);
__m128d b = load_m128d(_b);
__m128d c = _mm_div_pd(a, b);
return validate_double(c, d0, d1);
#else
return TEST_UNIMPL;
// #endif // ENABLE_TEST_ALL
#endif // ENABLE_TEST_ALL
}

result_t test_mm_div_sd(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
// #ifdef ENABLE_TEST_ALL
// const double *_a = (const double *)impl.test_cases_float_pointer1;
// const double *_b = (const double *)impl.test_cases_float_pointer2;
//
// double d0 = _a[0] / _b[0];
// double d1 = _a[1];
//
// __m128d a = load_m128d(_a);
// __m128d b = load_m128d(_b);
//
// __m128d c = _mm_div_sd(a, b);
//
// return validate_double(c, d0, d1);
// #else
#ifdef ENABLE_TEST_ALL
const double *_a = (const double *)impl.test_cases_float_pointer1;
const double *_b = (const double *)impl.test_cases_float_pointer2;

double d0 = _a[0] / _b[0];
double d1 = _a[1];

__m128d a = load_m128d(_a);
__m128d b = load_m128d(_b);

__m128d c = _mm_div_sd(a, b);

return validate_double(c, d0, d1);
#else
return TEST_UNIMPL;
// #endif // ENABLE_TEST_ALL
#endif // ENABLE_TEST_ALL
}

result_t test_mm_extract_epi16(const SSE2RVV_TEST_IMPL &impl, uint32_t iter) {
Expand Down

0 comments on commit 9b6ebcf

Please sign in to comment.