Skip to content

Fix float16_to_float(vint4) decoding every negative half as NaN on F16C builds #668

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

Open
lenamonj wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Fix float16_to_float(vint4) decoding every negative half as NaN on F16C builds #668

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
10 changes: 10 additions & 0 deletions Source/UnitTest/test_simd.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1154,6 +1154,16 @@ TEST(SuiteVfloat4, float16_to_float)

// NaNs must be preserved
EXPECT_NE(std::isnan(r.lane<3>()), 0);

// Negative numbers must keep their sign
vint4 b(0xBE00, 0xDB54, 0xFC00, 0x8000);
vfloat4 s = float16_to_float(b);

EXPECT_EQ(s.lane<0>(), -1.5);
EXPECT_EQ(s.lane<1>(), -234.5);
EXPECT_NE(std::isinf(s.lane<2>()), 0);
EXPECT_LT(s.lane<2>(), 0.0f);
EXPECT_NE(std::signbit(s.lane<3>()), 0);
}

/** @brief Test fp16 to float conversion. */
Expand Down
2 changes: 1 addition & 1 deletion Source/astcenc_vecmathlib_sse_4.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -998,7 +998,7 @@ static inline uint16_t float_to_float16(float a)
ASTCENC_SIMD_INLINE vfloat4 float16_to_float(vint4 a)
{
#if ASTCENC_F16C >= 1
__m128i packed = _mm_packs_epi32(a.m, a.m);
__m128i packed = _mm_packus_epi32(a.m, a.m);
__m128 f32 = _mm_cvtph_ps(packed);
return vfloat4(f32);
#else
Expand Down