@cpp-linter/clang-tidy-win32-x64 21.1.0 → 23.1.3

package/lib/clang/21/include/__clang_cuda_complex_builtins.h

@@ -1,285 +0,0 @@
-/*===-- __clang_cuda_complex_builtins - CUDA impls of runtime complex fns ---===
- *
- * Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
- * See https://llvm.org/LICENSE.txt for license information.
- * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
- *
- *===-----------------------------------------------------------------------===
- */
-
-#ifndef __CLANG_CUDA_COMPLEX_BUILTINS
-#define __CLANG_CUDA_COMPLEX_BUILTINS
-
-// This header defines __muldc3, __mulsc3, __divdc3, and __divsc3.  These are
-// libgcc functions that clang assumes are available when compiling c99 complex
-// operations.  (These implementations come from libc++, and have been modified
-// to work with CUDA and OpenMP target offloading [in C and C++ mode].)
-
-#pragma push_macro("__DEVICE__")
-#if defined(__OPENMP_NVPTX__) || defined(__OPENMP_AMDGCN__)
-#pragma omp declare target
-#define __DEVICE__ __attribute__((noinline, nothrow, cold, weak))
-#else
-#define __DEVICE__ __device__ inline
-#endif
-
-// To make the algorithms available for C and C++ in CUDA and OpenMP we select
-// different but equivalent function versions. TODO: For OpenMP we currently
-// select the native builtins as the overload support for templates is lacking.
-#if !defined(__OPENMP_NVPTX__) && !defined(__OPENMP_AMDGCN__)
-#define _ISNANd std::isnan
-#define _ISNANf std::isnan
-#define _ISINFd std::isinf
-#define _ISINFf std::isinf
-#define _ISFINITEd std::isfinite
-#define _ISFINITEf std::isfinite
-#define _COPYSIGNd std::copysign
-#define _COPYSIGNf std::copysign
-#define _SCALBNd std::scalbn
-#define _SCALBNf std::scalbn
-#define _ABSd std::abs
-#define _ABSf std::abs
-#define _LOGBd std::logb
-#define _LOGBf std::logb
-// Rather than pulling in std::max from algorithm everytime, use available ::max.
-#define _fmaxd max
-#define _fmaxf max
-#else
-#ifdef __AMDGCN__
-#define _ISNANd __ocml_isnan_f64
-#define _ISNANf __ocml_isnan_f32
-#define _ISINFd __ocml_isinf_f64
-#define _ISINFf __ocml_isinf_f32
-#define _ISFINITEd __ocml_isfinite_f64
-#define _ISFINITEf __ocml_isfinite_f32
-#define _COPYSIGNd __ocml_copysign_f64
-#define _COPYSIGNf __ocml_copysign_f32
-#define _SCALBNd __ocml_scalbn_f64
-#define _SCALBNf __ocml_scalbn_f32
-#define _ABSd __ocml_fabs_f64
-#define _ABSf __ocml_fabs_f32
-#define _LOGBd __ocml_logb_f64
-#define _LOGBf __ocml_logb_f32
-#define _fmaxd __ocml_fmax_f64
-#define _fmaxf __ocml_fmax_f32
-#else
-#define _ISNANd __nv_isnand
-#define _ISNANf __nv_isnanf
-#define _ISINFd __nv_isinfd
-#define _ISINFf __nv_isinff
-#define _ISFINITEd __nv_isfinited
-#define _ISFINITEf __nv_finitef
-#define _COPYSIGNd __nv_copysign
-#define _COPYSIGNf __nv_copysignf
-#define _SCALBNd __nv_scalbn
-#define _SCALBNf __nv_scalbnf
-#define _ABSd __nv_fabs
-#define _ABSf __nv_fabsf
-#define _LOGBd __nv_logb
-#define _LOGBf __nv_logbf
-#define _fmaxd __nv_fmax
-#define _fmaxf __nv_fmaxf
-#endif
-#endif
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-__DEVICE__ double _Complex __muldc3(double __a, double __b, double __c,
-                                    double __d) {
-  double __ac = __a * __c;
-  double __bd = __b * __d;
-  double __ad = __a * __d;
-  double __bc = __b * __c;
-  double _Complex z;
-  __real__(z) = __ac - __bd;
-  __imag__(z) = __ad + __bc;
-  if (_ISNANd(__real__(z)) && _ISNANd(__imag__(z))) {
-    int __recalc = 0;
-    if (_ISINFd(__a) || _ISINFd(__b)) {
-      __a = _COPYSIGNd(_ISINFd(__a) ? 1 : 0, __a);
-      __b = _COPYSIGNd(_ISINFd(__b) ? 1 : 0, __b);
-      if (_ISNANd(__c))
-        __c = _COPYSIGNd(0, __c);
-      if (_ISNANd(__d))
-        __d = _COPYSIGNd(0, __d);
-      __recalc = 1;
-    }
-    if (_ISINFd(__c) || _ISINFd(__d)) {
-      __c = _COPYSIGNd(_ISINFd(__c) ? 1 : 0, __c);
-      __d = _COPYSIGNd(_ISINFd(__d) ? 1 : 0, __d);
-      if (_ISNANd(__a))
-        __a = _COPYSIGNd(0, __a);
-      if (_ISNANd(__b))
-        __b = _COPYSIGNd(0, __b);
-      __recalc = 1;
-    }
-    if (!__recalc &&
-        (_ISINFd(__ac) || _ISINFd(__bd) || _ISINFd(__ad) || _ISINFd(__bc))) {
-      if (_ISNANd(__a))
-        __a = _COPYSIGNd(0, __a);
-      if (_ISNANd(__b))
-        __b = _COPYSIGNd(0, __b);
-      if (_ISNANd(__c))
-        __c = _COPYSIGNd(0, __c);
-      if (_ISNANd(__d))
-        __d = _COPYSIGNd(0, __d);
-      __recalc = 1;
-    }
-    if (__recalc) {
-      // Can't use std::numeric_limits<double>::infinity() -- that doesn't have
-      // a device overload (and isn't constexpr before C++11, naturally).
-      __real__(z) = __builtin_huge_val() * (__a * __c - __b * __d);
-      __imag__(z) = __builtin_huge_val() * (__a * __d + __b * __c);
-    }
-  }
-  return z;
-}
-
-__DEVICE__ float _Complex __mulsc3(float __a, float __b, float __c, float __d) {
-  float __ac = __a * __c;
-  float __bd = __b * __d;
-  float __ad = __a * __d;
-  float __bc = __b * __c;
-  float _Complex z;
-  __real__(z) = __ac - __bd;
-  __imag__(z) = __ad + __bc;
-  if (_ISNANf(__real__(z)) && _ISNANf(__imag__(z))) {
-    int __recalc = 0;
-    if (_ISINFf(__a) || _ISINFf(__b)) {
-      __a = _COPYSIGNf(_ISINFf(__a) ? 1 : 0, __a);
-      __b = _COPYSIGNf(_ISINFf(__b) ? 1 : 0, __b);
-      if (_ISNANf(__c))
-        __c = _COPYSIGNf(0, __c);
-      if (_ISNANf(__d))
-        __d = _COPYSIGNf(0, __d);
-      __recalc = 1;
-    }
-    if (_ISINFf(__c) || _ISINFf(__d)) {
-      __c = _COPYSIGNf(_ISINFf(__c) ? 1 : 0, __c);
-      __d = _COPYSIGNf(_ISINFf(__d) ? 1 : 0, __d);
-      if (_ISNANf(__a))
-        __a = _COPYSIGNf(0, __a);
-      if (_ISNANf(__b))
-        __b = _COPYSIGNf(0, __b);
-      __recalc = 1;
-    }
-    if (!__recalc &&
-        (_ISINFf(__ac) || _ISINFf(__bd) || _ISINFf(__ad) || _ISINFf(__bc))) {
-      if (_ISNANf(__a))
-        __a = _COPYSIGNf(0, __a);
-      if (_ISNANf(__b))
-        __b = _COPYSIGNf(0, __b);
-      if (_ISNANf(__c))
-        __c = _COPYSIGNf(0, __c);
-      if (_ISNANf(__d))
-        __d = _COPYSIGNf(0, __d);
-      __recalc = 1;
-    }
-    if (__recalc) {
-      __real__(z) = __builtin_huge_valf() * (__a * __c - __b * __d);
-      __imag__(z) = __builtin_huge_valf() * (__a * __d + __b * __c);
-    }
-  }
-  return z;
-}
-
-__DEVICE__ double _Complex __divdc3(double __a, double __b, double __c,
-                                    double __d) {
-  int __ilogbw = 0;
-  // Can't use std::max, because that's defined in <algorithm>, and we don't
-  // want to pull that in for every compile.  The CUDA headers define
-  // ::max(float, float) and ::max(double, double), which is sufficient for us.
-  double __logbw = _LOGBd(_fmaxd(_ABSd(__c), _ABSd(__d)));
-  if (_ISFINITEd(__logbw)) {
-    __ilogbw = (int)__logbw;
-    __c = _SCALBNd(__c, -__ilogbw);
-    __d = _SCALBNd(__d, -__ilogbw);
-  }
-  double __denom = __c * __c + __d * __d;
-  double _Complex z;
-  __real__(z) = _SCALBNd((__a * __c + __b * __d) / __denom, -__ilogbw);
-  __imag__(z) = _SCALBNd((__b * __c - __a * __d) / __denom, -__ilogbw);
-  if (_ISNANd(__real__(z)) && _ISNANd(__imag__(z))) {
-    if ((__denom == 0.0) && (!_ISNANd(__a) || !_ISNANd(__b))) {
-      __real__(z) = _COPYSIGNd(__builtin_huge_val(), __c) * __a;
-      __imag__(z) = _COPYSIGNd(__builtin_huge_val(), __c) * __b;
-    } else if ((_ISINFd(__a) || _ISINFd(__b)) && _ISFINITEd(__c) &&
-               _ISFINITEd(__d)) {
-      __a = _COPYSIGNd(_ISINFd(__a) ? 1.0 : 0.0, __a);
-      __b = _COPYSIGNd(_ISINFd(__b) ? 1.0 : 0.0, __b);
-      __real__(z) = __builtin_huge_val() * (__a * __c + __b * __d);
-      __imag__(z) = __builtin_huge_val() * (__b * __c - __a * __d);
-    } else if (_ISINFd(__logbw) && __logbw > 0.0 && _ISFINITEd(__a) &&
-               _ISFINITEd(__b)) {
-      __c = _COPYSIGNd(_ISINFd(__c) ? 1.0 : 0.0, __c);
-      __d = _COPYSIGNd(_ISINFd(__d) ? 1.0 : 0.0, __d);
-      __real__(z) = 0.0 * (__a * __c + __b * __d);
-      __imag__(z) = 0.0 * (__b * __c - __a * __d);
-    }
-  }
-  return z;
-}
-
-__DEVICE__ float _Complex __divsc3(float __a, float __b, float __c, float __d) {
-  int __ilogbw = 0;
-  float __logbw = _LOGBf(_fmaxf(_ABSf(__c), _ABSf(__d)));
-  if (_ISFINITEf(__logbw)) {
-    __ilogbw = (int)__logbw;
-    __c = _SCALBNf(__c, -__ilogbw);
-    __d = _SCALBNf(__d, -__ilogbw);
-  }
-  float __denom = __c * __c + __d * __d;
-  float _Complex z;
-  __real__(z) = _SCALBNf((__a * __c + __b * __d) / __denom, -__ilogbw);
-  __imag__(z) = _SCALBNf((__b * __c - __a * __d) / __denom, -__ilogbw);
-  if (_ISNANf(__real__(z)) && _ISNANf(__imag__(z))) {
-    if ((__denom == 0) && (!_ISNANf(__a) || !_ISNANf(__b))) {
-      __real__(z) = _COPYSIGNf(__builtin_huge_valf(), __c) * __a;
-      __imag__(z) = _COPYSIGNf(__builtin_huge_valf(), __c) * __b;
-    } else if ((_ISINFf(__a) || _ISINFf(__b)) && _ISFINITEf(__c) &&
-               _ISFINITEf(__d)) {
-      __a = _COPYSIGNf(_ISINFf(__a) ? 1 : 0, __a);
-      __b = _COPYSIGNf(_ISINFf(__b) ? 1 : 0, __b);
-      __real__(z) = __builtin_huge_valf() * (__a * __c + __b * __d);
-      __imag__(z) = __builtin_huge_valf() * (__b * __c - __a * __d);
-    } else if (_ISINFf(__logbw) && __logbw > 0 && _ISFINITEf(__a) &&
-               _ISFINITEf(__b)) {
-      __c = _COPYSIGNf(_ISINFf(__c) ? 1 : 0, __c);
-      __d = _COPYSIGNf(_ISINFf(__d) ? 1 : 0, __d);
-      __real__(z) = 0 * (__a * __c + __b * __d);
-      __imag__(z) = 0 * (__b * __c - __a * __d);
-    }
-  }
-  return z;
-}
-
-#if defined(__cplusplus)
-} // extern "C"
-#endif
-
-#undef _ISNANd
-#undef _ISNANf
-#undef _ISINFd
-#undef _ISINFf
-#undef _COPYSIGNd
-#undef _COPYSIGNf
-#undef _ISFINITEd
-#undef _ISFINITEf
-#undef _SCALBNd
-#undef _SCALBNf
-#undef _ABSd
-#undef _ABSf
-#undef _LOGBd
-#undef _LOGBf
-#undef _fmaxd
-#undef _fmaxf
-
-#if defined(__OPENMP_NVPTX__) || defined(__OPENMP_AMDGCN__)
-#pragma omp end declare target
-#endif
-
-#pragma pop_macro("__DEVICE__")
-
-#endif // __CLANG_CUDA_COMPLEX_BUILTINS

package/lib/clang/21/include/amxbf16transposeintrin.h

@@ -1,94 +0,0 @@
-/*===----- amxbf16transposeintrin.h - AMX-BF16 and AMX-TRANSPOSE ------------===
- *
- * Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
- * See https://llvm.org/LICENSE.txt for license information.
- * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
- *
- *===------------------------------------------------------------------------===
- */
-
-#ifndef __IMMINTRIN_H
-#error                                                                         \
-    "Never use <amxbf16transposeintrin.h> directly; use <immintrin.h> instead."
-#endif /* __IMMINTRIN_H */
-
-#ifndef __AMX_BF16TRANSPOSEINTRIN_H
-#define __AMX_BF16TRANSPOSEINTRIN_H
-#ifdef __x86_64__
-
-/* Define the default attributes for the functions in this file. */
-#define __DEFAULT_FN_ATTRS                                                     \
-  __attribute__((__always_inline__, __nodebug__,                               \
-                 __target__("amx-bf16,amx-transpose")))
-
-/// Compute transpose and dot-product of BF16 (16-bit) floating-point pairs in
-///    tiles \a a and \a b, accumulating the intermediate single-precision
-///    (32-bit) floating-point elements with elements in \a dst, and store the
-///    32-bit result back to tile \a dst.
-///
-/// \headerfile <immintrin.h>
-///
-/// \code
-/// void _tile_tdpbf16ps (__tile dst, __tile a, __tile b)
-/// \endcode
-///
-/// \code{.operation}
-/// FOR m := 0 TO dst.rows - 1
-///	tmp := dst.row[m]
-///	FOR k := 0 TO (a.colsb / 4) - 1
-///		FOR n := 0 TO (dst.colsb / 4) - 1
-///			tmp.bf32[n] += FP32(a.row[m].bf16[2*k+0]) *
-///					FP32(b.row[k].bf16[2*n+0])
-///			tmp.bf32[n] += FP32(a.row[m].bf16[2*k+1]) *
-///					FP32(b.row[k].bf16[2*n+1])
-///		ENDFOR
-///	ENDFOR
-///	write_row_and_zero(dst, m, tmp, dst.colsb)
-/// ENDFOR
-/// zero_upper_rows(dst, dst.rows)
-/// zero_tileconfig_start()
-/// \endcode
-///
-/// This intrinsic corresponds to the \c TTDPBF16PS instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param a
-///    The 1st source tile. Max size is 1024 Bytes.
-/// \param b
-///    The 2nd source tile. Max size is 1024 Bytes.
-#define _tile_tdpbf16ps(dst, a, b) __builtin_ia32_ttdpbf16ps((dst), (a), (b))
-
-/// This is internal intrinsic. C/C++ user should avoid calling it directly.
-static __inline__ _tile1024i __DEFAULT_FN_ATTRS
-_tile_tdpbf16ps_internal(unsigned short m, unsigned short n, unsigned short k,
-                         _tile1024i dst, _tile1024i src1, _tile1024i src2) {
-  return __builtin_ia32_ttdpbf16ps_internal(m, n, k, dst, src1, src2);
-}
-
-/// Compute transpose and dot-product of BF16 (16-bit) floating-point pairs in
-///    tiles src0 and src1, accumulating the intermediate single-precision
-///    (32-bit) floating-point elements with elements in "dst", and store the
-///    32-bit result back to tile "dst".
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> TTDPBF16PS </c> instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param src0
-///    The 1st source tile. Max size is 1024 Bytes.
-/// \param src1
-///    The 2nd source tile. Max size is 1024 Bytes.
-__DEFAULT_FN_ATTRS
-static __inline__ void __tile_tdpbf16ps(__tile1024i *dst, __tile1024i src0,
-                                        __tile1024i src1) {
-  dst->tile = _tile_tdpbf16ps_internal(src0.row, src1.col, src0.col, dst->tile,
-                                       src0.tile, src1.tile);
-}
-
-#undef __DEFAULT_FN_ATTRS
-
-#endif /* __x86_64__ */
-#endif /* __AMX_BF16TRANSPOSEINTRIN_H */

package/lib/clang/21/include/amxcomplextransposeintrin.h

@@ -1,303 +0,0 @@
-/*===----- amxcomplextransposeintrin.h - AMX-COMPLEX and AMX-TRANSPOSE ------===
- *
- * Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
- * See https://llvm.org/LICENSE.txt for license information.
- * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
- *
- *===------------------------------------------------------------------------===
- */
-
-#ifndef __IMMINTRIN_H
-#error                                                                         \
-    "Never use <amxcomplextransposeintrin.h> directly; include <immintrin.h> instead."
-#endif // __IMMINTRIN_H
-
-#ifndef __AMX_COMPLEXTRANSPOSEINTRIN_H
-#define __AMX_COMPLEXTRANSPOSEINTRIN_H
-#ifdef __x86_64__
-
-#define __DEFAULT_FN_ATTRS                                                     \
-  __attribute__((__always_inline__, __nodebug__,                               \
-                 __target__("amx-complex,amx-transpose")))
-
-/// Perform matrix multiplication of two tiles containing complex elements and
-///    accumulate the results into a packed single precision tile. Each dword
-///    element in input tiles \a a and \a b is interpreted as a complex number
-///    with FP16 real part and FP16 imaginary part.
-/// Calculates the imaginary part of the result. For each possible combination
-///    of (transposed column of \a a, column of \a b), it performs a set of
-///    multiplication and accumulations on all corresponding complex numbers
-///    (one from \a a and one from \a b). The imaginary part of the \a a element
-///    is multiplied with the real part of the corresponding \a b element, and
-///    the real part of the \a a element is multiplied with the imaginary part
-///    of the corresponding \a b elements. The two accumulated results are
-///    added, and then accumulated into the corresponding row and column of
-///    \a dst.
-///
-/// \headerfile <x86intrin.h>
-///
-/// \code
-/// void _tile_tcmmimfp16ps(__tile dst, __tile a, __tile b);
-/// \endcode
-///
-/// \code{.operation}
-/// FOR m := 0 TO dst.rows - 1
-///	tmp := dst.row[m]
-///	FOR k := 0 TO a.rows - 1
-///		FOR n := 0 TO (dst.colsb / 4) - 1
-///			tmp.fp32[n] += FP32(a.row[m].fp16[2*k+0]) * FP32(b.row[k].fp16[2*n+1])
-///			tmp.fp32[n] += FP32(a.row[m].fp16[2*k+1]) * FP32(b.row[k].fp16[2*n+0])
-///		ENDFOR
-///	ENDFOR
-///	write_row_and_zero(dst, m, tmp, dst.colsb)
-/// ENDFOR
-/// zero_upper_rows(dst, dst.rows)
-/// zero_tileconfig_start()
-/// \endcode
-///
-/// This intrinsic corresponds to the \c TTCMMIMFP16PS instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param a
-///    The 1st source tile. Max size is 1024 Bytes.
-/// \param b
-///    The 2nd source tile. Max size is 1024 Bytes.
-#define _tile_tcmmimfp16ps(dst, a, b)                                          \
-  __builtin_ia32_ttcmmimfp16ps((dst), (a), (b))
-
-/// Perform matrix multiplication of two tiles containing complex elements and
-///    accumulate the results into a packed single precision tile. Each dword
-///    element in input tiles \a a and \a b is interpreted as a complex number
-///    with FP16 real part and FP16 imaginary part.
-/// Calculates the real part of the result. For each possible combination
-///    of (rtransposed colum of \a a, column of \a b), it performs a set of
-///    multiplication and accumulations on all corresponding complex numbers
-///    (one from \a a and one from \a b). The real part of the \a a element is
-///    multiplied with the real part of the corresponding \a b element, and the
-///    negated imaginary part of the \a a element is multiplied with the
-///    imaginary part of the corresponding \a b elements. The two accumulated
-///    results are added, and then accumulated into the corresponding row and
-///    column of \a dst.
-///
-/// \headerfile <x86intrin.h>
-///
-/// \code
-/// void _tile_tcmmrlfp16ps(__tile dst, __tile a, __tile b);
-/// \endcode
-///
-/// \code{.operation}
-/// FOR m := 0 TO dst.rows - 1
-///	tmp := dst.row[m]
-///	FOR k := 0 TO a.rows - 1
-///		FOR n := 0 TO (dst.colsb / 4) - 1
-///			tmp.fp32[n] += FP32(a.row[m].fp16[2*k+0]) * FP32(b.row[k].fp16[2*n+0])
-///			tmp.fp32[n] += FP32(-a.row[m].fp16[2*k+1]) * FP32(b.row[k].fp16[2*n+1])
-///		ENDFOR
-///	ENDFOR
-///	write_row_and_zero(dst, m, tmp, dst.colsb)
-/// ENDFOR
-/// zero_upper_rows(dst, dst.rows)
-/// zero_tileconfig_start()
-/// \endcode
-///
-/// This intrinsic corresponds to the \c TTCMMIMFP16PS instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param a
-///    The 1st source tile. Max size is 1024 Bytes.
-/// \param b
-///    The 2nd source tile. Max size is 1024 Bytes.
-#define _tile_tcmmrlfp16ps(dst, a, b)                                          \
-  __builtin_ia32_ttcmmrlfp16ps((dst), (a), (b))
-
-/// Perform matrix conjugate transpose and multiplication of two tiles
-///    containing complex elements and accumulate the results into a packed
-///    single precision tile. Each dword element in input tiles \a a and \a b
-///    is interpreted as a complex number with FP16 real part and FP16 imaginary
-///    part.
-/// Calculates the imaginary part of the result. For each possible combination
-///    of (transposed column of \a a, column of \a b), it performs a set of
-///    multiplication and accumulations on all corresponding complex numbers
-///    (one from \a a and one from \a b). The negated imaginary part of the \a a
-///    element is multiplied with the real part of the corresponding \a b
-///    element, and the real part of the \a a element is multiplied with the
-///    imaginary part of the corresponding \a b elements. The two accumulated
-///    results are added, and then accumulated into the corresponding row and
-///    column of \a dst.
-///
-/// \headerfile <x86intrin.h>
-///
-/// \code
-/// void _tile_conjtcmmimfp16ps(__tile dst, __tile a, __tile b);
-/// \endcode
-///
-/// \code{.operation}
-/// FOR m := 0 TO dst.rows - 1
-///	tmp := dst.row[m]
-///	FOR k := 0 TO a.rows - 1
-///		FOR n := 0 TO (dst.colsb / 4) - 1
-///			tmp.fp32[n] += FP32(a.row[m].fp16[2*k+0]) * FP32(b.row[k].fp16[2*n+1])
-///			tmp.fp32[n] += FP32(-a.row[m].fp16[2*k+1]) * FP32(b.row[k].fp16[2*n+0])
-///		ENDFOR
-///	ENDFOR
-///	write_row_and_zero(dst, m, tmp, dst.colsb)
-/// ENDFOR
-/// zero_upper_rows(dst, dst.rows)
-/// zero_tileconfig_start()
-/// \endcode
-///
-/// This intrinsic corresponds to the \c TCONJTCMMIMFP16PS instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param a
-///    The 1st source tile. Max size is 1024 Bytes.
-/// \param b
-///    The 2nd source tile. Max size is 1024 Bytes.
-#define _tile_conjtcmmimfp16ps(dst, a, b)                                      \
-  __builtin_ia32_tconjtcmmimfp16ps((dst), (a), (b))
-
-/// Perform conjugate transpose of an FP16-pair of complex elements from \a a
-///    and writes the result to \a dst.
-///
-/// \headerfile <x86intrin.h>
-///
-/// \code
-/// void _tile_conjtfp16(__tile dst, __tile a);
-/// \endcode
-///
-/// \code{.operation}
-/// FOR i := 0 TO dst.rows - 1
-///	FOR j := 0 TO (dst.colsb / 4) - 1
-///		tmp.fp16[2*j+0] := a.row[j].fp16[2*i+0]
-///		tmp.fp16[2*j+1] := -a.row[j].fp16[2*i+1]
-///	ENDFOR
-///	write_row_and_zero(dst, i, tmp, dst.colsb)
-/// ENDFOR
-/// zero_upper_rows(dst, dst.rows)
-/// zero_tileconfig_start()
-/// \endcode
-///
-/// This intrinsic corresponds to the \c TCONJTFP16 instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param a
-///    The source tile. Max size is 1024 Bytes.
-#define _tile_conjtfp16(dst, a) __builtin_ia32_tconjtfp16((dst), (a))
-
-static __inline__ _tile1024i __DEFAULT_FN_ATTRS _tile_tcmmimfp16ps_internal(
-    unsigned short m, unsigned short n, unsigned short k, _tile1024i dst,
-    _tile1024i src1, _tile1024i src2) {
-  return __builtin_ia32_ttcmmimfp16ps_internal(m, n, k, dst, src1, src2);
-}
-
-static __inline__ _tile1024i __DEFAULT_FN_ATTRS _tile_tcmmrlfp16ps_internal(
-    unsigned short m, unsigned short n, unsigned short k, _tile1024i dst,
-    _tile1024i src1, _tile1024i src2) {
-  return __builtin_ia32_ttcmmrlfp16ps_internal(m, n, k, dst, src1, src2);
-}
-
-static __inline__ _tile1024i __DEFAULT_FN_ATTRS _tile_conjtcmmimfp16ps_internal(
-    unsigned short m, unsigned short n, unsigned short k, _tile1024i dst,
-    _tile1024i src1, _tile1024i src2) {
-  return __builtin_ia32_tconjtcmmimfp16ps_internal(m, n, k, dst, src1, src2);
-}
-
-static __inline__ _tile1024i __DEFAULT_FN_ATTRS
-_tile_conjtfp16_internal(unsigned short m, unsigned short n, _tile1024i src) {
-  return __builtin_ia32_tconjtfp16_internal(m, n, src);
-}
-
-/// Perform matrix multiplication of two tiles containing complex elements and
-///    accumulate the results into a packed single precision tile. Each dword
-///    element in input tiles src0 and src1 is interpreted as a complex number
-///    with FP16 real part and FP16 imaginary part.
-///    This function calculates the imaginary part of the result.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> TTCMMIMFP16PS </c> instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param src0
-///    The 1st source tile. Max size is 1024 Bytes.
-/// \param src1
-///    The 2nd source tile. Max size is 1024 Bytes.
-__DEFAULT_FN_ATTRS
-static void __tile_tcmmimfp16ps(__tile1024i *dst, __tile1024i src0,
-                                __tile1024i src1) {
-  dst->tile = _tile_tcmmimfp16ps_internal(src0.row, src1.col, src0.col,
-                                          dst->tile, src0.tile, src1.tile);
-}
-
-/// Perform matrix multiplication of two tiles containing complex elements and
-///    accumulate the results into a packed single precision tile. Each dword
-///    element in input tiles src0 and src1 is interpreted as a complex number
-///    with FP16 real part and FP16 imaginary part.
-///    This function calculates the real part of the result.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> TTCMMRLFP16PS </c> instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param src0
-///    The 1st source tile. Max size is 1024 Bytes.
-/// \param src1
-///    The 2nd source tile. Max size is 1024 Bytes.
-__DEFAULT_FN_ATTRS
-static void __tile_tcmmrlfp16ps(__tile1024i *dst, __tile1024i src0,
-                                __tile1024i src1) {
-  dst->tile = _tile_tcmmrlfp16ps_internal(src0.row, src1.col, src0.col,
-                                          dst->tile, src0.tile, src1.tile);
-}
-
-/// Perform matrix conjugate transpose and multiplication of two tiles
-///    containing complex elements and accumulate the results into a packed
-///    single precision tile. Each dword element in input tiles src0 and src1
-///    is interpreted as a complex number with FP16 real part and FP16 imaginary
-///    part.
-///    This function calculates the imaginary part of the result.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> TCONJTCMMIMFP16PS </c> instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param src0
-///    The 1st source tile. Max size is 1024 Bytes.
-/// \param src1
-///    The 2nd source tile. Max size is 1024 Bytes.
-__DEFAULT_FN_ATTRS
-static void __tile_conjtcmmimfp16ps(__tile1024i *dst, __tile1024i src0,
-                                    __tile1024i src1) {
-  dst->tile = _tile_conjtcmmimfp16ps_internal(src0.row, src1.col, src0.col,
-                                              dst->tile, src0.tile, src1.tile);
-}
-
-/// Perform conjugate transpose of an FP16-pair of complex elements from src and
-///    writes the result to dst.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> TCONJTFP16 </c> instruction.
-///
-/// \param dst
-///    The destination tile. Max size is 1024 Bytes.
-/// \param src
-///    The source tile. Max size is 1024 Bytes.
-__DEFAULT_FN_ATTRS
-static void __tile_conjtfp16(__tile1024i *dst, __tile1024i src) {
-  dst->tile = _tile_conjtfp16_internal(src.row, src.col, src.tile);
-}
-
-#undef __DEFAULT_FN_ATTRS
-
-#endif // __x86_64__
-#endif // __AMX_COMPLEXTRANSPOSEINTRIN_H