diff options
| author | Thiago Macieira <thiago.macieira@intel.com> | 2017-11-09 15:14:31 -0800 |
|---|---|---|
| committer | Thiago Macieira <thiago.macieira@intel.com> | 2017-12-15 07:39:54 +0000 |
| commit | 259acdcb8a38b8989a137942d20ed2685a4c940a (patch) | |
| tree | e529816eef15763ddc7c4df7e26b5e6820fc05a9 /src/corelib/global/qfloat16_f16c.c | |
| parent | 0b4b864d83b18c18f4e743754ccbc1134713f352 (diff) | |
qfloat16_f16c.c: inline in qfloat16.cpp if F16C is permitted in there
This is going to be the most common scenario for GCC: all recent
versions allow compiling F16C code in the same source, thus generating
better code.
MSVC is excluded unless AVX is already turned on by the user in the
mkspec file, because it fails to use the VEX-prefixed instructions for
everything else, printing a warning that it knows it should have done
so.
ICC is excluded because it's known to generate invalid code when using
the F16C intrinsics unless F16C is turned explicitly on.
Clang is excluded because it runs into an internal error compiling this
code unless F16C is turned explicitly on.
Change-Id: I57a1bd6e0c194530b732fffd14f58de6a46f0de1
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Diffstat (limited to 'src/corelib/global/qfloat16_f16c.c')
| -rw-r--r-- | src/corelib/global/qfloat16_f16c.c | 4 |
1 files changed, 3 insertions, 1 deletions
diff --git a/src/corelib/global/qfloat16_f16c.c b/src/corelib/global/qfloat16_f16c.c index ffd35f8ebc..c88dbb6944 100644 --- a/src/corelib/global/qfloat16_f16c.c +++ b/src/corelib/global/qfloat16_f16c.c @@ -44,7 +44,7 @@ // have been compiled if the support was missing in the first place, and not // all compilers define it. Technically, we didn't need to check for __AVX__ // either. -#if !defined(__AVX__) +#if !QT_COMPILER_SUPPORTS_HERE(AVX) # error "AVX support required" #endif @@ -53,6 +53,7 @@ QT_BEGIN_NAMESPACE extern "C" { #endif +QT_FUNCTION_TARGET(F16C) void qFloatToFloat16_fast(quint16 *out, const float *in, qssize_t len) Q_DECL_NOTHROW { qssize_t i = 0; @@ -67,6 +68,7 @@ void qFloatToFloat16_fast(quint16 *out, const float *in, qssize_t len) Q_DECL_NO out[i] = _mm_extract_epi16(_mm_cvtps_ph(_mm_set_ss(in[i]), 0), 0); } +QT_FUNCTION_TARGET(F16C) void qFloatFromFloat16_fast(float *out, const quint16 *in, qssize_t len) Q_DECL_NOTHROW { qssize_t i = 0; |