Index: lib/Headers/__emmintrin_f16c.h
===================================================================
--- lib/Headers/__emmintrin_f16c.h
+++ lib/Headers/__emmintrin_f16c.h
@@ -1,4 +1,4 @@
-/*===---- f16cintrin.h - F16C intrinsics -----------------------------------===
+/*===---- __emmintrin_f16c.h - F16C intrinsics -----------------------------===
  *
  * Permission is hereby granted, free of charge, to any person obtaining a copy
  * of this software and associated documentation files (the "Software"), to deal
@@ -21,12 +21,12 @@
  *===-----------------------------------------------------------------------===
  */
 
-#if !defined __X86INTRIN_H && !defined __EMMINTRIN_H && !defined __IMMINTRIN_H
-#error "Never use <f16cintrin.h> directly; include <emmintrin.h> instead."
+#if !defined __EMMINTRIN_H
+#error "Never use <__emmintrin_f16c.h> directly; include <emmintrin.h> instead."
 #endif
 
-#ifndef __F16CINTRIN_H
-#define __F16CINTRIN_H
+#ifndef __EMMINTRIN_F16C_H
+#define __EMMINTRIN_F16C_H
 
 /* Define the default attributes for the functions in this file. */
 #define __DEFAULT_FN_ATTRS \
@@ -121,4 +121,4 @@
 
 #undef __DEFAULT_FN_ATTRS
 
-#endif /* __F16CINTRIN_H */
+#endif /* __EMMINTRIN_F16C_H */
Index: lib/Headers/emmintrin.h
===================================================================
--- lib/Headers/emmintrin.h
+++ lib/Headers/emmintrin.h
@@ -44,7 +44,7 @@
  * appear in the interface though. */
 typedef signed char __v16qs __attribute__((__vector_size__(16)));
 
-#include <f16cintrin.h>
+#include <__emmintrin_f16c.h>
 
 /* Define the default attributes for the functions in this file. */
 #define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__, __target__("sse2")))
Index: lib/Headers/f16cintrin.h
===================================================================
--- lib/Headers/f16cintrin.h
+++ lib/Headers/f16cintrin.h
@@ -21,8 +21,8 @@
  *===-----------------------------------------------------------------------===
  */
 
-#if !defined __X86INTRIN_H && !defined __EMMINTRIN_H && !defined __IMMINTRIN_H
-#error "Never use <f16cintrin.h> directly; include <emmintrin.h> instead."
+#if !defined __IMMINTRIN_H
+#error "Never use <f16cintrin.h> directly; include <immintrin.h> instead."
 #endif
 
 #ifndef __F16CINTRIN_H
@@ -32,63 +32,24 @@
 #define __DEFAULT_FN_ATTRS \
   __attribute__((__always_inline__, __nodebug__, __target__("f16c")))
 
-/// Converts a 16-bit half-precision float value into a 32-bit float
-///    value.
-///
-/// \headerfile <x86intrin.h>
-///
-/// This intrinsic corresponds to the <c> VCVTPH2PS </c> instruction.
-///
-/// \param __a
-///    A 16-bit half-precision float value.
-/// \returns The converted 32-bit float value.
-static __inline float __DEFAULT_FN_ATTRS
-_cvtsh_ss(unsigned short __a)
-{
-  __v8hi v = {(short)__a, 0, 0, 0, 0, 0, 0, 0};
-  __v4sf r = __builtin_ia32_vcvtph2ps(v);
-  return r[0];
-}
-
-/// Converts a 32-bit single-precision float value to a 16-bit
-///    half-precision float value.
-///
-/// \headerfile <x86intrin.h>
-///
-/// \code
-/// unsigned short _cvtss_sh(float a, const int imm);
-/// \endcode
-///
-/// This intrinsic corresponds to the <c> VCVTPS2PH </c> instruction.
-///
-/// \param a
-///    A 32-bit single-precision float value to be converted to a 16-bit
-///    half-precision float value.
-/// \param imm
-///    An immediate value controlling rounding using bits [2:0]: \n
-///    000: Nearest \n
-///    001: Down \n
-///    010: Up \n
-///    011: Truncate \n
-///    1XX: Use MXCSR.RC for rounding
-/// \returns The converted 16-bit half-precision float value.
-#define _cvtss_sh(a, imm) __extension__ ({ \
-  (unsigned short)(((__v8hi)__builtin_ia32_vcvtps2ph((__v4sf){a, 0, 0, 0}, \
-                                                     (imm)))[0]); })
+/* The 256-bit versions of functions in f16cintrin.h.
+   Intel documents these as being in immintrin.h, and
+   they depend on typedefs from avxintrin.h. */
 
-/// Converts a 128-bit vector containing 32-bit float values into a
-///    128-bit vector containing 16-bit half-precision float values.
+/// Converts a 256-bit vector of [8 x float] into a 128-bit vector
+///    containing 16-bit half-precision float values.
 ///
 /// \headerfile <x86intrin.h>
 ///
 /// \code
-/// __m128i _mm_cvtps_ph(__m128 a, const int imm);
+/// __m128i _mm256_cvtps_ph(__m256 a, const int imm);
 /// \endcode
 ///
 /// This intrinsic corresponds to the <c> VCVTPS2PH </c> instruction.
 ///
 /// \param a
-///    A 128-bit vector containing 32-bit float values.
+///    A 256-bit vector containing 32-bit single-precision float values to be
+///    converted to 16-bit half-precision float values.
 /// \param imm
 ///    An immediate value controlling rounding using bits [2:0]: \n
 ///    000: Nearest \n
@@ -96,27 +57,27 @@
 ///    010: Up \n
 ///    011: Truncate \n
 ///    1XX: Use MXCSR.RC for rounding
-/// \returns A 128-bit vector containing converted 16-bit half-precision float
-///    values. The lower 64 bits are used to store the converted 16-bit
-///    half-precision floating-point values.
-#define _mm_cvtps_ph(a, imm) __extension__ ({ \
-  (__m128i)__builtin_ia32_vcvtps2ph((__v4sf)(__m128)(a), (imm)); })
+/// \returns A 128-bit vector containing the converted 16-bit half-precision
+///    float values.
+#define _mm256_cvtps_ph(a, imm) __extension__ ({ \
+ (__m128i)__builtin_ia32_vcvtps2ph256((__v8sf)(__m256)(a), (imm)); })
 
 /// Converts a 128-bit vector containing 16-bit half-precision float
-///    values into a 128-bit vector containing 32-bit float values.
+///    values into a 256-bit vector of [8 x float].
 ///
 /// \headerfile <x86intrin.h>
 ///
 /// This intrinsic corresponds to the <c> VCVTPH2PS </c> instruction.
 ///
 /// \param __a
-///    A 128-bit vector containing 16-bit half-precision float values. The lower
-///    64 bits are used in the conversion.
-/// \returns A 128-bit vector of [4 x float] containing converted float values.
-static __inline __m128 __DEFAULT_FN_ATTRS
-_mm_cvtph_ps(__m128i __a)
+///    A 128-bit vector containing 16-bit half-precision float values to be
+///    converted to 32-bit single-precision float values.
+/// \returns A vector of [8 x float] containing the converted 32-bit
+///    single-precision float values.
+static __inline __m256 __attribute__((__always_inline__, __nodebug__, __target__("f16c")))
+_mm256_cvtph_ps(__m128i __a)
 {
-  return (__m128)__builtin_ia32_vcvtph2ps((__v8hi)__a);
+  return (__m256)__builtin_ia32_vcvtph2ps256((__v8hi)__a);
 }
 
 #undef __DEFAULT_FN_ATTRS
Index: lib/Headers/immintrin.h
===================================================================
--- lib/Headers/immintrin.h
+++ lib/Headers/immintrin.h
@@ -69,54 +69,8 @@
 #if !defined(_MSC_VER) || __has_feature(modules) || defined(__AVX2__)
 #include <avx2intrin.h>
 
-/* The 256-bit versions of functions in f16cintrin.h.
-   Intel documents these as being in immintrin.h, and
-   they depend on typedefs from avxintrin.h. */
-
-/// Converts a 256-bit vector of [8 x float] into a 128-bit vector
-///    containing 16-bit half-precision float values.
-///
-/// \headerfile <x86intrin.h>
-///
-/// \code
-/// __m128i _mm256_cvtps_ph(__m256 a, const int imm);
-/// \endcode
-///
-/// This intrinsic corresponds to the <c> VCVTPS2PH </c> instruction.
-///
-/// \param a
-///    A 256-bit vector containing 32-bit single-precision float values to be
-///    converted to 16-bit half-precision float values.
-/// \param imm
-///    An immediate value controlling rounding using bits [2:0]: \n
-///    000: Nearest \n
-///    001: Down \n
-///    010: Up \n
-///    011: Truncate \n
-///    1XX: Use MXCSR.RC for rounding
-/// \returns A 128-bit vector containing the converted 16-bit half-precision
-///    float values.
-#define _mm256_cvtps_ph(a, imm) __extension__ ({ \
- (__m128i)__builtin_ia32_vcvtps2ph256((__v8sf)(__m256)(a), (imm)); })
-
-/// Converts a 128-bit vector containing 16-bit half-precision float
-///    values into a 256-bit vector of [8 x float].
-///
-/// \headerfile <x86intrin.h>
-///
-/// This intrinsic corresponds to the <c> VCVTPH2PS </c> instruction.
-///
-/// \param __a
-///    A 128-bit vector containing 16-bit half-precision float values to be
-///    converted to 32-bit single-precision float values.
-/// \returns A vector of [8 x float] containing the converted 32-bit
-///    single-precision float values.
-static __inline __m256 __attribute__((__always_inline__, __nodebug__, __target__("f16c")))
-_mm256_cvtph_ps(__m128i __a)
-{
-  return (__m256)__builtin_ia32_vcvtph2ps256((__v8hi)__a);
-}
-#endif /* __AVX2__ */
+#if !defined(_MSC_VER) || __has_feature(modules) || defined(__F16C__)
+#include <f16cintrin.h>
 
 #if !defined(_MSC_VER) || __has_feature(modules) || defined(__VPCLMULQDQ__)
 #include <vpclmulqdqintrin.h>