Skip to content

x86 avx: fix cvtt bounds #1393

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
mr-c merged 1 commit into from
Mar 4, 2026
+60 −18
Merged

x86 avx: fix cvtt bounds #1393

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
38 changes: 26 additions & 12 deletions simde/x86/avx.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3823,7 +3823,7 @@ simde__m128i
simde_mm256_cvttpd_epi32 (simde__m256d a) {
#if defined(SIMDE_X86_AVX_NATIVE)
return _mm256_cvttpd_epi32(a);
#elif defined(SIMDE_LOONGARCH_LASX_NATIVE)
#elif defined(SIMDE_LOONGARCH_LASX_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE)
simde__m256i_private a_;
a_.i256 = __lasx_xvftintrz_w_d(a, a);
a_.i256 = __lasx_xvpermi_d(a_.i256, 0xd8);
Expand All @@ -3832,13 +3832,20 @@ simde_mm256_cvttpd_epi32 (simde__m256d a) {
simde__m128i_private r_;
simde__m256d_private a_ = simde__m256d_to_private(a);

#if defined(simde_math_trunc)
#if SIMDE_NATURAL_VECTOR_SIZE_LE(128)
r_.m64[0] = simde_mm_cvttpd_pi32(a_.m128d[0]);
r_.m64[1] = simde_mm_cvttpd_pi32(a_.m128d[1]);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(a_.f64) / sizeof(a_.f64[0])) ; i++) {
r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, simde_math_trunc(a_.f64[i]));
for (size_t i = 0 ; i < (sizeof(r_.i32) / sizeof(r_.i32[0])) ; i++) {
simde_float64 v = simde_math_trunc(a_.f64[i]);
#if defined(SIMDE_FAST_CONVERSION_RANGE)
r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
#else
r_.i32[i] = ((v > HEDLEY_STATIC_CAST(simde_float64, INT32_MIN)) && (v < HEDLEY_STATIC_CAST(simde_float64, INT32_MAX))) ?
SIMDE_CONVERT_FTOI(int32_t, v) : INT32_MIN;
#endif
}
#else
HEDLEY_UNREACHABLE();
#endif

return simde__m128i_from_private(r_);
Expand All @@ -3854,19 +3861,26 @@ simde__m256i
simde_mm256_cvttps_epi32 (simde__m256 a) {
#if defined(SIMDE_X86_AVX_NATIVE)
return _mm256_cvttps_epi32(a);
#elif defined(SIMDE_LOONGARCH_LASX_NATIVE)
#elif defined(SIMDE_LOONGARCH_LASX_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE)
return __lasx_xvftintrz_w_s(a);
#else
simde__m256i_private r_;
simde__m256_private a_ = simde__m256_to_private(a);

#if defined(simde_math_truncf)
#if SIMDE_NATURAL_VECTOR_SIZE_LE(128)
r_.m128i[0] = simde_mm_cvttps_epi32(a_.m128[0]);
r_.m128i[1] = simde_mm_cvttps_epi32(a_.m128[1]);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(a_.f32) / sizeof(a_.f32[0])) ; i++) {
r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, simde_math_truncf(a_.f32[i]));
for (size_t i = 0 ; i < (sizeof(r_.i32) / sizeof(r_.i32[0])) ; i++) {
simde_float32 v = simde_math_truncf(a_.f32[i]);
#if defined(SIMDE_FAST_CONVERSION_RANGE)
r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
#else
r_.i32[i] = ((v > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) && (v < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX))) ?
SIMDE_CONVERT_FTOI(int32_t, v) : INT32_MIN;
#endif
}
#else
HEDLEY_UNREACHABLE();
#endif

return simde__m256i_from_private(r_);
Expand Down
40 changes: 34 additions & 6 deletions test/x86/avx.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7125,9 +7125,9 @@ test_simde_mm256_cvtpd_epi32(SIMDE_MUNIT_TEST_ARGS) {
#endif
#if !defined(SIMDE_FAST_CONVERSION_RANGE)
{ simde_mm256_set_pd(
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) + 1),
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) + 1),
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) - 100),
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) - 1),
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) - 1),
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) + 100)),
simde_mm_set_epi32(
INT32_MIN, INT32_C(2147483547), INT32_MIN, -INT32_C(2147483548)) },
Expand Down Expand Up @@ -7218,9 +7218,9 @@ test_simde_mm256_cvtps_epi32(SIMDE_MUNIT_TEST_ARGS) {
#endif
#if !defined(SIMDE_FAST_CONVERSION_RANGE)
{ simde_mm256_set_ps(
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) + 1),
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) + 1),
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) - 100),
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) - 1),
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) - 1),
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) + 100),
0.f, 0.f, 0.f, 0.f),
simde_mm256_set_epi32(
Expand Down Expand Up @@ -7436,7 +7436,20 @@ test_simde_mm256_cvttpd_epi32(SIMDE_MUNIT_TEST_ARGS) {
const struct {
simde__m256d a;
simde__m128i r;
} test_vec[8] = {
} test_vec[] = {
#if !defined(SIMDE_FAST_NANS)
{ simde_mm256_set_pd(SIMDE_MATH_NAN, -SIMDE_MATH_NAN, 0.0, 0.0),
simde_mm_set_epi32( INT32_MIN, INT32_MIN, 0, 0) },
#endif
#if !defined(SIMDE_FAST_CONVERSION_RANGE)
{ simde_mm256_set_pd(
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) + 1),
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) - 100),
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) - 1),
HEDLEY_STATIC_CAST(simde_float64, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) + 100)),
simde_mm_set_epi32(
INT32_MIN, INT32_C(2147483547), INT32_MIN, -INT32_C(2147483548)) },
#endif
{ simde_mm256_set_pd(SIMDE_FLOAT64_C( -175.82), SIMDE_FLOAT64_C( -91.19),
SIMDE_FLOAT64_C( -855.64), SIMDE_FLOAT64_C(-1000.00)),
simde_mm_set_epi32(INT32_C(-175), INT32_C( -91), INT32_C(-855), INT32_C(-1000)) },
Expand Down Expand Up @@ -7476,7 +7489,22 @@ test_simde_mm256_cvttps_epi32(SIMDE_MUNIT_TEST_ARGS) {
const struct {
simde__m256 a;
simde__m256i r;
} test_vec[8] = {
} test_vec[] = {
#if !defined(SIMDE_FAST_NANS)
{ simde_mm256_set_ps(SIMDE_MATH_NAN, -SIMDE_MATH_NAN, 0.f, 0.f, 0.f, 0.f, 0.f, 0.f),
simde_mm256_set_epi32( INT32_MIN, INT32_MIN, 0, 0, 0, 0, 0, 0) },
#endif
#if !defined(SIMDE_FAST_CONVERSION_RANGE)
{ simde_mm256_set_ps(
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) + 1),
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MAX) - 100),
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) - 1),
HEDLEY_STATIC_CAST(simde_float32, HEDLEY_STATIC_CAST(int64_t, INT32_MIN) + 100),
0.f, 0.f, 0.f, 0.f),
simde_mm256_set_epi32(
INT32_MIN, INT32_C(2147483520), INT32_MIN, -INT32_C(2147483520),
0, 0, 0, 0) },
#endif
{ simde_mm256_set_ps(SIMDE_FLOAT32_C( -135.75), SIMDE_FLOAT32_C( 534.39),
SIMDE_FLOAT32_C( -81.93), SIMDE_FLOAT32_C( -234.94),
SIMDE_FLOAT32_C( -390.94), SIMDE_FLOAT32_C( -625.05),
Expand Down