cannam@167: /*
cannam@167:  * Copyright (c) 2003, 2007-14 Matteo Frigo
cannam@167:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
cannam@167:  *
cannam@167:  * This program is free software; you can redistribute it and/or modify
cannam@167:  * it under the terms of the GNU General Public License as published by
cannam@167:  * the Free Software Foundation; either version 2 of the License, or
cannam@167:  * (at your option) any later version.
cannam@167:  *
cannam@167:  * This program is distributed in the hope that it will be useful,
cannam@167:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
cannam@167:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
cannam@167:  * GNU General Public License for more details.
cannam@167:  *
cannam@167:  * You should have received a copy of the GNU General Public License
cannam@167:  * along with this program; if not, write to the Free Software
cannam@167:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
cannam@167:  *
cannam@167:  */
cannam@167: 
cannam@167: /* detection of alignment.  This is complicated because a machine may
cannam@167:    support multiple SIMD extensions (e.g. SSE2 and AVX) but only one
cannam@167:    set of alignment contraints.  So this alignment stuff cannot be
cannam@167:    defined in the SIMD header files.  Rather than defining a separate
cannam@167:    set of "machine" header files, we just do this ugly ifdef here. */
cannam@167: #if defined(HAVE_AVX512)
cannam@167: #  if defined(FFTW_SINGLE)
cannam@167: #    define ALIGNMENT 8     /* Alignment for the LD/ST macros */
cannam@167: #    define ALIGNMENTA 64   /* Alignment for the LDA/STA macros */
cannam@167: #  else
cannam@167: #    define ALIGNMENT 16    /* Alignment for the LD/ST macros */
cannam@167: #    define ALIGNMENTA 64   /* Alignment for the LDA/STA macros */
cannam@167: #  endif
cannam@167: #elif defined(HAVE_SSE2) || defined(HAVE_AVX) || defined(HAVE_AVX2) || (HAVE_AVX_128_FMA)
cannam@167: #  if defined(FFTW_SINGLE)
cannam@167: #    define ALIGNMENT 8     /* Alignment for the LD/ST macros */
cannam@167: #    define ALIGNMENTA 16   /* Alignment for the LDA/STA macros */
cannam@167: #  else
cannam@167: #    define ALIGNMENT 16    /* Alignment for the LD/ST macros */
cannam@167: #    define ALIGNMENTA 16   /* Alignment for the LDA/STA macros */
cannam@167: #  endif
cannam@167: #elif defined(HAVE_ALTIVEC)
cannam@167: #  define ALIGNMENT 8     /* Alignment for the LD/ST macros */
cannam@167: #  define ALIGNMENTA 16   /* Alignment for the LDA/STA macros */
cannam@167: #elif defined(HAVE_NEON) || defined(HAVE_VSX)
cannam@167: #  define ALIGNMENT 8     /* Alignment for the LD/ST macros */
cannam@167: #  define ALIGNMENTA 8    /* Alignment for the LDA/STA macros */
cannam@167: #elif defined(HAVE_KCVI)
cannam@167: #  if defined(FFTW_SINGLE)
cannam@167: #    define ALIGNMENT 8     /* Alignment for the LD/ST macros */
cannam@167: #  else
cannam@167: #    define ALIGNMENT 16     /* Alignment for the LD/ST macros */
cannam@167: #  endif
cannam@167: #  define ALIGNMENTA 64   /* Alignment for the LDA/STA macros */
cannam@167: #elif defined(HAVE_GENERIC_SIMD256)
cannam@167: #  if defined(FFTW_SINGLE)
cannam@167: #    define ALIGNMENT 8
cannam@167: #    define ALIGNMENTA 32
cannam@167: #  else
cannam@167: #    define ALIGNMENT 16
cannam@167: #    define ALIGNMENTA 32
cannam@167: #  endif
cannam@167: #elif defined(HAVE_GENERIC_SIMD128)
cannam@167: #  if defined(FFTW_SINGLE)
cannam@167: #    define ALIGNMENT 8
cannam@167: #    define ALIGNMENTA 16
cannam@167: #  else
cannam@167: #    define ALIGNMENT 16
cannam@167: #    define ALIGNMENTA 16
cannam@167: #  endif
cannam@167: #endif
cannam@167: 
cannam@167: #if HAVE_SIMD
cannam@167: #  ifndef ALIGNMENT
cannam@167: #  error "ALIGNMENT not defined"
cannam@167: #  endif
cannam@167: #  ifndef ALIGNMENTA
cannam@167: #  error "ALIGNMENTA not defined"
cannam@167: #  endif
cannam@167: #endif
cannam@167: 
cannam@167: /* rename for precision and for SIMD extensions */
cannam@167: #define XSIMD0(name, suffix) CONCAT(name, suffix)
cannam@167: #define XSIMD(name) XSIMD0(X(name), SIMD_SUFFIX)
cannam@167: #define XSIMD_STRING(x) x STRINGIZE(SIMD_SUFFIX)
cannam@167: 
cannam@167: /* TAINT_BIT is set if pointers are not guaranteed to be multiples of
cannam@167:    ALIGNMENT */
cannam@167: #define TAINT_BIT 1    
cannam@167: 
cannam@167: /* TAINT_BITA is set if pointers are not guaranteed to be multiples of
cannam@167:    ALIGNMENTA */
cannam@167: #define TAINT_BITA 2
cannam@167: 
cannam@167: #define PTRINT(p) ((uintptr_t)(p))
cannam@167: 
cannam@167: #define ALIGNED(p) \
cannam@167:   (((PTRINT(UNTAINT(p)) % ALIGNMENT) == 0) && !(PTRINT(p) & TAINT_BIT))
cannam@167: 
cannam@167: #define ALIGNEDA(p) \
cannam@167:   (((PTRINT(UNTAINT(p)) % ALIGNMENTA) == 0) && !(PTRINT(p) & TAINT_BITA))
cannam@167: 
cannam@167: #define SIMD_STRIDE_OK(x) (!(((x) * sizeof(R)) % ALIGNMENT))
cannam@167: #define SIMD_STRIDE_OKA(x) (!(((x) * sizeof(R)) % ALIGNMENTA))
cannam@167: #define SIMD_VSTRIDE_OK SIMD_STRIDE_OK
cannam@167: