@zigc/lib 0.17.0-dev.76 → 0.17.0-dev.813

package/include/amxmovrstransposeintrin.h

@@ -1,200 +0,0 @@
-/* ===--- amxmovrstransposeintrin.h - AMX_MOVRS_TRANSPOSE intrinsics --------===
- *
- * 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 <amxmovrstransposeintrin.h> directly; use <immintrin.h> instead."
-#endif /* __IMMINTRIN_H */
-
-#ifndef __AMX_MOVRS_TRANSPOSEINTRIN_H
-#define __AMX_MOVRS_TRANSPOSEINTRIN_H
-#ifdef __x86_64__
-
-#define __DEFAULT_FN_ATTRS                                                     \
-  __attribute__((__always_inline__, __nodebug__,                               \
-                 __target__("amx-transpose,amx-movrs")))
-
-#define _tile_2rpntlvwz0rs(tdst, base, stride)                                 \
-  __builtin_ia32_t2rpntlvwz0rs(tdst, base, stride)
-#define _tile_2rpntlvwz0rst1(tdst, base, stride)                               \
-  __builtin_ia32_t2rpntlvwz0rst1(tdst, base, stride)
-#define _tile_2rpntlvwz1rs(tdst, base, stride)                                 \
-  __builtin_ia32_t2rpntlvwz1rs(tdst, base, stride)
-#define _tile_2rpntlvwz1rst1(tdst, base, stride)                               \
-  __builtin_ia32_t2rpntlvwz1rst1(tdst, base, stride)
-
-static __inline__ void __DEFAULT_FN_ATTRS _tile_2rpntlvwz0rs_internal(
-    unsigned short row, unsigned short col0, unsigned short col1,
-    _tile1024i *dst0, _tile1024i *dst1, const void *base,
-    __SIZE_TYPE__ stride) {
-  // Use __tile1024i_1024a* to escape the alignment check in
-  // clang/test/Headers/x86-intrinsics-headers-clean.cpp
-  __builtin_ia32_t2rpntlvwz0rs_internal(
-      row, col0, col1, (_tile1024i_1024a *)dst0, (_tile1024i_1024a *)dst1, base,
-      (__SIZE_TYPE__)(stride));
-}
-
-static __inline__ void __DEFAULT_FN_ATTRS _tile_2rpntlvwz0rst1_internal(
-    unsigned short row, unsigned short col0, unsigned short col1,
-    _tile1024i *dst0, _tile1024i *dst1, const void *base,
-    __SIZE_TYPE__ stride) {
-  __builtin_ia32_t2rpntlvwz0rst1_internal(
-      row, col0, col1, (_tile1024i_1024a *)dst0, (_tile1024i_1024a *)dst1, base,
-      (__SIZE_TYPE__)(stride));
-}
-
-static __inline__ void __DEFAULT_FN_ATTRS _tile_2rpntlvwz1rs_internal(
-    unsigned short row, unsigned short col0, unsigned short col1,
-    _tile1024i *dst0, _tile1024i *dst1, const void *base,
-    __SIZE_TYPE__ stride) {
-  __builtin_ia32_t2rpntlvwz1rs_internal(
-      row, col0, col1, (_tile1024i_1024a *)dst0, (_tile1024i_1024a *)dst1, base,
-      (__SIZE_TYPE__)(stride));
-}
-
-static __inline__ void __DEFAULT_FN_ATTRS _tile_2rpntlvwz1rst1_internal(
-    unsigned short row, unsigned short col0, unsigned short col1,
-    _tile1024i *dst0, _tile1024i *dst1, const void *base,
-    __SIZE_TYPE__ stride) {
-  __builtin_ia32_t2rpntlvwz1rst1_internal(
-      row, col0, col1, (_tile1024i_1024a *)dst0, (_tile1024i_1024a *)dst1, base,
-      (__SIZE_TYPE__)(stride));
-}
-
-/// Converts a pair of tiles from memory into VNNI format, and places the
-/// results in a pair of destinations specified by dst. The pair of tiles
-/// in memory is specified via a tsib; the second tile is after the first
-/// one, separated by the same stride that separates each row.
-/// The tile configuration for the destination tiles indicates the amount
-/// of data to read from memory. The instruction will load a number of rows
-/// that is equal to twice the number of rows in tmm1. The size of each row
-/// is equal to the average width of the destination tiles. If the second
-/// tile is configured with zero rows and columns, only the first tile will
-/// be written.
-/// Provides a hint to the implementation that the data will likely become
-/// read shared in the near future and the data caching can be optimized.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> T2RPNTLVWZ0RS </c> instruction.
-///
-/// \param dst0
-///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param dst1
-///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param base
-///    A pointer to base address.
-/// \param stride
-///    The stride between the rows' data to be loaded in memory.
-__DEFAULT_FN_ATTRS
-static void __tile_2rpntlvwz0rs(__tile1024i *dst0, __tile1024i *dst1,
-                                const void *base, __SIZE_TYPE__ stride) {
-  _tile_2rpntlvwz0rs_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
-                              &dst1->tile, base, stride);
-}
-
-/// Converts a pair of tiles from memory into VNNI format, and places the
-/// results in a pair of destinations specified by dst. The pair of tiles
-/// in memory is specified via a tsib; the second tile is after the first
-/// one, separated by the same stride that separates each row.
-/// The tile configuration for the destination tiles indicates the amount
-/// of data to read from memory. The instruction will load a number of rows
-/// that is equal to twice the number of rows in tmm1. The size of each row
-/// is equal to the average width of the destination tiles. If the second
-/// tile is configured with zero rows and columns, only the first tile will
-/// be written.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> T2RPNTLVWZ0T1RS </c> instruction.
-///
-/// \param dst0
-///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param dst1
-///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param base
-///    A pointer to base address.
-/// \param stride
-///    The stride between the rows' data to be loaded in memory.
-__DEFAULT_FN_ATTRS
-static void __tile_2rpntlvwz0rst1(__tile1024i *dst0, __tile1024i *dst1,
-                                  const void *base, __SIZE_TYPE__ stride) {
-  _tile_2rpntlvwz0rst1_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
-                                &dst1->tile, base, stride);
-}
-
-/// Converts a pair of tiles from memory into VNNI format, and places the
-/// results in a pair of destinations specified by dst. The pair of tiles
-/// in memory is specified via a tsib; the second tile is after the first
-/// one, separated by the same stride that separates each row.
-/// The tile configuration for the destination tiles indicates the amount
-/// of data to read from memory. The instruction will load a number of rows
-/// that is equal to twice the number of rows in tmm1. The size of each row
-/// is equal to the average width of the destination tiles. If the second
-/// tile is configured with zero rows and columns, only the first tile will
-/// be written. The last row will be not be read from memory but instead
-/// filled with zeros.
-/// Provides a hint to the implementation that the data will likely become
-/// read shared in the near future and the data caching can be optimized.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> T2RPNTLVWZ1 </c> instruction.
-///
-/// \param dst0
-///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param dst1
-///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param base
-///    A pointer to base address.
-/// \param stride
-///    The stride between the rows' data to be loaded in memory.
-__DEFAULT_FN_ATTRS
-static void __tile_2rpntlvwz1rs(__tile1024i *dst0, __tile1024i *dst1,
-                                const void *base, __SIZE_TYPE__ stride) {
-  _tile_2rpntlvwz1rs_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
-                              &dst1->tile, base, stride);
-}
-
-/// Converts a pair of tiles from memory into VNNI format, and places the
-/// results in a pair of destinations specified by dst. The pair of tiles
-/// in memory is specified via a tsib; the second tile is after the first
-/// one, separated by the same stride that separates each row.
-/// The tile configuration for the destination tiles indicates the amount
-/// of data to read from memory. The instruction will load a number of rows
-/// that is equal to twice the number of rows in tmm1. The size of each row
-/// is equal to the average width of the destination tiles. If the second
-/// tile is configured with zero rows and columns, only the first tile will
-/// be written. The last row will be not be read from memory but instead
-/// filled with zeros.
-/// Provides a hint to the implementation that the data will likely become
-/// read shared in the near future and the data caching can be optimized.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> T2RPNTLVWZ1T1RS </c> instruction.
-///
-/// \param dst0
-///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param dst1
-///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param base
-///    A pointer to base address.
-/// \param stride
-///    The stride between the rows' data to be loaded in memory.
-__DEFAULT_FN_ATTRS
-static void __tile_2rpntlvwz1rst1(__tile1024i *dst0, __tile1024i *dst1,
-                                  const void *base, __SIZE_TYPE__ stride) {
-  _tile_2rpntlvwz1rst1_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
-                                &dst1->tile, base, stride);
-}
-
-#undef __DEFAULT_FN_ATTRS
-#endif /* __x86_64__ */
-#endif /* __AMX_MOVRS_TRANSPOSEINTRIN_H */

package/include/amxtf32transposeintrin.h

@@ -1,105 +0,0 @@
-/*===--------- amxtf32transposeintrin.h - AMX-TF32 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 <amxtf32transposeintrin.h> directly; include <immintrin.h> instead."
-#endif // __IMMINTRIN_H
-
-#ifndef __AMX_TF32TRANSPOSEINTRIN_H
-#define __AMX_TF32TRANSPOSEINTRIN_H
-#ifdef __x86_64__
-
-#define __DEFAULT_FN_ATTRS_TF32_TRANSPOSE                                      \
-  __attribute__((__always_inline__, __nodebug__,                               \
-                 __target__("amx-tf32,amx-transpose")))
-
-/// \code
-/// void _tile_tmmultf32ps(constexpr int srcdst, constexpr int a, \
-///                        constexpr int b);
-/// \endcode
-///
-/// This intrinsic corresponds to the <c> TTMMULTF32PS </c> instruction.
-///
-/// \param srcdst
-/// 	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.
-///
-/// \code{.operation}
-/// DEFINE zero_lower_mantissa_bits_fp32(x[31:0]) {
-/// 	dword[12:0] := 0
-/// 	dword[31:13] := x[31:13]
-/// 	return dword
-/// }
-///
-/// DEFINE silence_snan_fp32(x[31:0]) {
-/// 	IF (x.exponent == 255 and x.fraction != 0 and x.fraction[22] == 0)
-/// 		x.fraction[22] := 1
-/// 	return x
-/// }
-///
-/// elements_dest:= srcdst.colsb/4
-///
-/// FOR m := 0 TO (srcdst.rows-1)
-/// 	tmp[511:0] := 0
-/// 	FOR k := 0 TO (a.rows-1)
-/// 		FOR n := 0 TO (elements_dest-1)
-/// 			a1e := silence_snan_fp32(a.row[k].fp32[m])
-/// 			a2e := silence_snan_fp32(b.row[k].fp32[n])
-/// 			s1e := zero_lower_mantissa_bits_fp32(a1e)
-/// 			s2e := zero_lower_mantissa_bits_fp32(a2e)
-/// 			tmp.fp32[n] += s1e * s2e
-/// 		ENDFOR
-/// 	ENDFOR
-///
-/// 	FOR n := 0 TO (elements_dest-1)
-/// 		tmp.fp32[n] += srcdst.row[m].fp32[n]
-/// 	ENDFOR
-///	write_row_and_zero(srcdst, m, tmp, srcdst.colsb)
-///
-/// ENDFOR
-///
-/// zero_upper_rows(srcdst, srcdst.rows)
-/// zero_tileconfig_start()
-/// \endcode
-#define _tile_tmmultf32ps(srcdst, a, b)                                        \
-  __builtin_ia32_ttmmultf32ps((srcdst), (a), (b))
-
-// dst = m x n (srcdest), src1 = k x m, src2 = k x n
-static __inline__ _tile1024i __DEFAULT_FN_ATTRS_TF32_TRANSPOSE
-_tile_tmmultf32ps_internal(unsigned short m, unsigned short n, unsigned short k,
-                           _tile1024i dst, _tile1024i src1, _tile1024i src2) {
-  return __builtin_ia32_ttmmultf32ps_internal(m, n, k, dst, src1, src2);
-}
-
-/// Compute transpose and do Matrix Multiplication of src0 and src1, and then do
-/// Matrix Plus with dst. All the calculation is base on float32 but with the
-/// lower 13-bit set to 0.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> TTMMULTF32PS </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_TF32_TRANSPOSE
-static void __tile_tmmultf32ps(__tile1024i *dst, __tile1024i src0,
-                               __tile1024i src1) {
-  dst->tile = _tile_tmmultf32ps_internal(src0.row, src1.col, src0.col,
-                                         dst->tile, src0.tile, src1.tile);
-}
-
-#endif // __x86_64__
-#endif // __AMX_TF32TRANSPOSEINTRIN_H

package/include/amxtransposeintrin.h

@@ -1,248 +0,0 @@
-/* ===--- amxtransposeintrin.h - AMX_TRANSPOSE intrinsics -*- C++ -*---------===
- *
- * 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 <amxtransposeintrin.h> directly; use <immintrin.h> instead."
-#endif /* __IMMINTRIN_H */
-
-#ifndef __AMX_TRANSPOSEINTRIN_H
-#define __AMX_TRANSPOSEINTRIN_H
-#ifdef __x86_64__
-
-#define __DEFAULT_FN_ATTRS_TRANSPOSE                                           \
-  __attribute__((__always_inline__, __nodebug__, __target__("amx-transpose")))
-
-#define _tile_2rpntlvwz0(tdst, base, stride)                                   \
-  __builtin_ia32_t2rpntlvwz0(tdst, base, stride)
-#define _tile_2rpntlvwz0t1(tdst, base, stride)                                 \
-  __builtin_ia32_t2rpntlvwz0t1(tdst, base, stride)
-#define _tile_2rpntlvwz1(tdst, base, stride)                                   \
-  __builtin_ia32_t2rpntlvwz1(tdst, base, stride)
-#define _tile_2rpntlvwz1t1(tdst, base, stride)                                 \
-  __builtin_ia32_t2rpntlvwz1t1(tdst, base, stride)
-
-/// Transpose 32-bit elements from \a src and write the result to \a dst.
-///
-/// \headerfile <immintrin.h>
-///
-/// \code
-/// void _tile_transposed(__tile dst, __tile src);
-/// \endcode
-///
-/// This intrinsic corresponds to the <c> TTRANSPOSED </c> instruction.
-///
-/// \param dst
-/// 	The destination tile. Max size is 1024 Bytes.
-/// \param src
-/// 	The source tile. Max size is 1024 Bytes.
-///
-/// \code{.operation}
-///
-/// FOR i := 0 TO (dst.rows-1)
-/// 	tmp[511:0] := 0
-/// 	FOR j := 0 TO (dst.colsb/4-1)
-/// 		tmp.dword[j] := src.row[j].dword[i]
-/// 	ENDFOR
-/// 	dst.row[i] := tmp
-/// ENDFOR
-///
-/// zero_upper_rows(dst, dst.rows)
-/// zero_tileconfig_start()
-/// \endcode
-#define _tile_transposed(dst, src) __builtin_ia32_ttransposed(dst, src)
-
-static __inline__ void __DEFAULT_FN_ATTRS_TRANSPOSE _tile_2rpntlvwz0_internal(
-    unsigned short row, unsigned short col0, unsigned short col1,
-    _tile1024i *dst0, _tile1024i *dst1, const void *base,
-    __SIZE_TYPE__ stride) {
-  // Use __tile1024i_1024a* to escape the alignment check in
-  // clang/test/Headers/x86-intrinsics-headers-clean.cpp
-  __builtin_ia32_t2rpntlvwz0_internal(row, col0, col1, (_tile1024i_1024a *)dst0,
-                                      (_tile1024i_1024a *)dst1, base,
-                                      (__SIZE_TYPE__)(stride));
-}
-
-static __inline__ void __DEFAULT_FN_ATTRS_TRANSPOSE _tile_2rpntlvwz0t1_internal(
-    unsigned short row, unsigned short col0, unsigned short col1,
-    _tile1024i *dst0, _tile1024i *dst1, const void *base,
-    __SIZE_TYPE__ stride) {
-  __builtin_ia32_t2rpntlvwz0t1_internal(
-      row, col0, col1, (_tile1024i_1024a *)dst0, (_tile1024i_1024a *)dst1, base,
-      (__SIZE_TYPE__)(stride));
-}
-
-static __inline__ void __DEFAULT_FN_ATTRS_TRANSPOSE _tile_2rpntlvwz1_internal(
-    unsigned short row, unsigned short col0, unsigned short col1,
-    _tile1024i *dst0, _tile1024i *dst1, const void *base,
-    __SIZE_TYPE__ stride) {
-  __builtin_ia32_t2rpntlvwz1_internal(row, col0, col1, (_tile1024i_1024a *)dst0,
-                                      (_tile1024i_1024a *)dst1, base,
-                                      (__SIZE_TYPE__)(stride));
-}
-
-static __inline__ void __DEFAULT_FN_ATTRS_TRANSPOSE _tile_2rpntlvwz1t1_internal(
-    unsigned short row, unsigned short col0, unsigned short col1,
-    _tile1024i *dst0, _tile1024i *dst1, const void *base,
-    __SIZE_TYPE__ stride) {
-  __builtin_ia32_t2rpntlvwz1t1_internal(
-      row, col0, col1, (_tile1024i_1024a *)dst0, (_tile1024i_1024a *)dst1, base,
-      (__SIZE_TYPE__)(stride));
-}
-
-// This is internal intrinsic. C/C++ user should avoid calling it directly.
-static __inline__ _tile1024i __DEFAULT_FN_ATTRS_TRANSPOSE
-_tile_transposed_internal(unsigned short m, unsigned short n, _tile1024i src) {
-  return __builtin_ia32_ttransposed_internal(m, n, src);
-}
-
-/// Converts a pair of tiles from memory into VNNI format, and places the
-/// results in a pair of destinations specified by dst. The pair of tiles
-/// in memory is specified via a tsib; the second tile is after the first
-/// one, separated by the same stride that separates each row.
-/// The tile configuration for the destination tiles indicates the amount
-/// of data to read from memory. The instruction will load a number of rows
-/// that is equal to twice the number of rows in tmm1. The size of each row
-/// is equal to the average width of the destination tiles. If the second
-/// tile is configured with zero rows and columns, only the first tile will
-/// be written.
-/// Provides a hint to the implementation that the data will likely not be
-/// reused in the near future and the data caching can be optimized.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> T2RPNTLVWZ0 </c> instruction.
-///
-/// \param dst0
-///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param dst1
-///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param base
-///    A pointer to base address.
-/// \param stride
-///    The stride between the rows' data to be loaded in memory.
-__DEFAULT_FN_ATTRS_TRANSPOSE
-static void __tile_2rpntlvwz0(__tile1024i *dst0, __tile1024i *dst1,
-                              const void *base, __SIZE_TYPE__ stride) {
-  _tile_2rpntlvwz0_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
-                            &dst1->tile, base, stride);
-}
-
-/// Converts a pair of tiles from memory into VNNI format, and places the
-/// results in a pair of destinations specified by dst. The pair of tiles
-/// in memory is specified via a tsib; the second tile is after the first
-/// one, separated by the same stride that separates each row.
-/// The tile configuration for the destination tiles indicates the amount
-/// of data to read from memory. The instruction will load a number of rows
-/// that is equal to twice the number of rows in tmm1. The size of each row
-/// is equal to the average width of the destination tiles. If the second
-/// tile is configured with zero rows and columns, only the first tile will
-/// be written.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> T2RPNTLVWZ0T1 </c> instruction.
-///
-/// \param dst0
-///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param dst1
-///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param base
-///    A pointer to base address.
-/// \param stride
-///    The stride between the rows' data to be loaded in memory.
-__DEFAULT_FN_ATTRS_TRANSPOSE
-static void __tile_2rpntlvwz0t1(__tile1024i *dst0, __tile1024i *dst1,
-                                const void *base, __SIZE_TYPE__ stride) {
-  _tile_2rpntlvwz0t1_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
-                              &dst1->tile, base, stride);
-}
-
-/// Converts a pair of tiles from memory into VNNI format, and places the
-/// results in a pair of destinations specified by dst. The pair of tiles
-/// in memory is specified via a tsib; the second tile is after the first
-/// one, separated by the same stride that separates each row.
-/// The tile configuration for the destination tiles indicates the amount
-/// of data to read from memory. The instruction will load a number of rows
-/// that is equal to twice the number of rows in tmm1. The size of each row
-/// is equal to the average width of the destination tiles. If the second
-/// tile is configured with zero rows and columns, only the first tile will
-/// be written. The last row will be not be read from memory but instead
-/// filled with zeros.
-/// Provides a hint to the implementation that the data will likely not be
-/// reused in the near future and the data caching can be optimized.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> T2RPNTLVWZ1 </c> instruction.
-///
-/// \param dst0
-///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param dst1
-///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param base
-///    A pointer to base address.
-/// \param stride
-///    The stride between the rows' data to be loaded in memory.
-__DEFAULT_FN_ATTRS_TRANSPOSE
-static void __tile_2rpntlvwz1(__tile1024i *dst0, __tile1024i *dst1,
-                              const void *base, __SIZE_TYPE__ stride) {
-  _tile_2rpntlvwz1_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
-                            &dst1->tile, base, stride);
-}
-
-/// Converts a pair of tiles from memory into VNNI format, and places the
-/// results in a pair of destinations specified by dst. The pair of tiles
-/// in memory is specified via a tsib; the second tile is after the first
-/// one, separated by the same stride that separates each row.
-/// The tile configuration for the destination tiles indicates the amount
-/// of data to read from memory. The instruction will load a number of rows
-/// that is equal to twice the number of rows in tmm1. The size of each row
-/// is equal to the average width of the destination tiles. If the second
-/// tile is configured with zero rows and columns, only the first tile will
-/// be written. The last row will be not be read from memory but instead
-/// filled with zeros.
-/// Provides a hint to the implementation that the data will likely not be
-/// reused in the near future and the data caching can be optimized.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> T2RPNTLVWZ1T1 </c> instruction.
-///
-/// \param dst0
-///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param dst1
-///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
-/// \param base
-///    A pointer to base address.
-/// \param stride
-///    The stride between the rows' data to be loaded in memory.
-__DEFAULT_FN_ATTRS_TRANSPOSE
-static void __tile_2rpntlvwz1t1(__tile1024i *dst0, __tile1024i *dst1,
-                                const void *base, __SIZE_TYPE__ stride) {
-  _tile_2rpntlvwz1t1_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
-                              &dst1->tile, base, stride);
-}
-
-/// Transpose 32-bit elements from src and write the result to dst.
-///
-/// \headerfile <immintrin.h>
-///
-/// This intrinsic corresponds to the <c> TTRANSPOSED </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_TRANSPOSE
-static void __tile_transposed(__tile1024i *dst, __tile1024i src) {
-  dst->tile = _tile_transposed_internal(dst->row, dst->col, src.tile);
-}
-
-#endif /* __x86_64__ */
-#endif /* __AMX_TRANSPOSEINTRIN_H */

package/libc/include/arm-netbsd-eabi/machine/frame.h

@@ -1,130 +0,0 @@
-/*	$NetBSD: frame.h,v 1.23 2022/04/02 11:16:07 skrll Exp $	*/
-
-/*
- * Copyright (c) 1994-1997 Mark Brinicombe.
- * Copyright (c) 1994 Brini.
- * All rights reserved.
- *
- * This code is derived from software written for Brini by Mark Brinicombe
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code must retain the above copyright
- *    notice, this list of conditions and the following disclaimer.
- * 2. Redistributions in binary form must reproduce the above copyright
- *    notice, this list of conditions and the following disclaimer in the
- *    documentation and/or other materials provided with the distribution.
- * 3. All advertising materials mentioning features or use of this software
- *    must display the following acknowledgement:
- *	This product includes software developed by Brini.
- * 4. The name of the company nor the name of the author may be used to
- *    endorse or promote products derived from this software without specific
- *    prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY BRINI ``AS IS'' AND ANY EXPRESS OR IMPLIED
- * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
- * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
- * IN NO EVENT SHALL BRINI OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT,
- * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
- * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
- * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
- * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
- * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
- * SUCH DAMAGE.
- */
-/*
- * arm/frame.h - Stack frames structures
- */
-
-#ifndef _ARM_FRAME_H_
-#define _ARM_FRAME_H_
-
-#ifndef _LOCORE
-
-#include <sys/signal.h>
-#include <sys/ucontext.h>
-
-/*
- * Trap frame.  Pushed onto the kernel stack on a trap (synchronous exception).
- */
-
-typedef struct trapframe {
-	register_t tf_spsr;
-	register_t tf_fill; /* fill here so r0 will be dword aligned */
-	register_t tf_r0;
-	register_t tf_r1;
-	register_t tf_r2;
-	register_t tf_r3;
-	register_t tf_r4;
-	register_t tf_r5;
-	register_t tf_r6;
-	register_t tf_r7;
-	register_t tf_r8;
-	register_t tf_r9;
-	register_t tf_r10;
-	register_t tf_r11;
-	register_t tf_r12;
-	register_t tf_usr_sp;
-	register_t tf_usr_lr;
-	register_t tf_svc_sp;
-	register_t tf_svc_lr;
-	register_t tf_pc;
-} trapframe_t;
-
-/* Register numbers */
-#define tf_ip tf_r12
-#define tf_r13 tf_usr_sp
-#define tf_r14 tf_usr_lr
-#define tf_r15 tf_pc
-
-#define TRAP_USERMODE(tf)	(((tf)->tf_spsr & PSR_MODE) == PSR_USR32_MODE)
-
-#define FB_R4	0
-#define FB_R5	1
-#define FB_R6	2
-#define FB_R7	3
-#define FB_R8	4
-#define FB_R9	5
-#define FB_R10	6
-#define FB_R11	7
-#define FB_R12	8
-#define FB_R13	9
-#define FB_R14	10
-#define FB_MAX	11
-struct faultbuf {
-	register_t fb_reg[FB_MAX];
-};
-
-/*
- * Signal frame.  Pushed onto user stack before calling sigcode.
- */
-#ifdef COMPAT_16
-struct sigframe_sigcontext {
-	struct	sigcontext sf_sc;
-};
-#endif
-
-/* the pointers are use in the trampoline code to locate the ucontext */
-struct sigframe_siginfo {
-	siginfo_t	sf_si;		/* actual saved siginfo */
-	ucontext_t	sf_uc;		/* actual saved ucontext */
-};
-
-#if defined(_KERNEL) || defined(_KMEMUSER)
-#ifdef _KERNEL
-__BEGIN_DECLS
-void sendsig_sigcontext(const ksiginfo_t *, const sigset_t *);
-void *getframe(struct lwp *, int, int *);
-__END_DECLS
-#define lwp_settrapframe(l, tf)		((l)->l_md.md_tf = (tf))
-#endif
-#define lwp_trapframe(l)		((l)->l_md.md_tf)
-#endif /* _KERNEL || _KMEMUSER */
-
-#endif /* _LOCORE */
-
-#endif /* _ARM_FRAME_H_ */
-
-/* End of frame.h */