mirror of
https://github.com/tildearrow/furnace.git
synced 2024-11-16 09:45:06 +00:00
331 lines
9.3 KiB
C
331 lines
9.3 KiB
C
/*
|
|
* Copyright (c) 2003, 2007-14 Matteo Frigo
|
|
* Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
|
|
*
|
|
* This program is free software; you can redistribute it and/or modify
|
|
* it under the terms of the GNU General Public License as published by
|
|
* the Free Software Foundation; either version 2 of the License, or
|
|
* (at your option) any later version.
|
|
*
|
|
* This program is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with this program; if not, write to the Free Software
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|
*
|
|
*/
|
|
|
|
#include "dft/codelet-dft.h"
|
|
#include SIMD_HEADER
|
|
|
|
#define EXTERN_CONST(t, x) extern const t x; const t x
|
|
|
|
static int n1b_okp(const kdft_desc *d,
|
|
const R *ri, const R *ii, const R *ro, const R *io,
|
|
INT is, INT os, INT vl, INT ivs, INT ovs,
|
|
const planner *plnr)
|
|
{
|
|
return (1
|
|
&& ALIGNED(ii)
|
|
&& ALIGNED(io)
|
|
&& !NO_SIMDP(plnr)
|
|
&& SIMD_STRIDE_OK(is)
|
|
&& SIMD_STRIDE_OK(os)
|
|
&& SIMD_VSTRIDE_OK(ivs)
|
|
&& SIMD_VSTRIDE_OK(ovs)
|
|
&& ri == ii + 1
|
|
&& ro == io + 1
|
|
&& (vl % VL) == 0
|
|
&& (!d->is || (d->is == is))
|
|
&& (!d->os || (d->os == os))
|
|
&& (!d->ivs || (d->ivs == ivs))
|
|
&& (!d->ovs || (d->ovs == ovs))
|
|
);
|
|
}
|
|
|
|
EXTERN_CONST(kdft_genus, XSIMD(dft_n1bsimd_genus)) = { n1b_okp, VL };
|
|
|
|
static int n1f_okp(const kdft_desc *d,
|
|
const R *ri, const R *ii, const R *ro, const R *io,
|
|
INT is, INT os, INT vl, INT ivs, INT ovs,
|
|
const planner *plnr)
|
|
{
|
|
return (1
|
|
&& ALIGNED(ri)
|
|
&& ALIGNED(ro)
|
|
&& !NO_SIMDP(plnr)
|
|
&& SIMD_STRIDE_OK(is)
|
|
&& SIMD_STRIDE_OK(os)
|
|
&& SIMD_VSTRIDE_OK(ivs)
|
|
&& SIMD_VSTRIDE_OK(ovs)
|
|
&& ii == ri + 1
|
|
&& io == ro + 1
|
|
&& (vl % VL) == 0
|
|
&& (!d->is || (d->is == is))
|
|
&& (!d->os || (d->os == os))
|
|
&& (!d->ivs || (d->ivs == ivs))
|
|
&& (!d->ovs || (d->ovs == ovs))
|
|
);
|
|
}
|
|
|
|
EXTERN_CONST(kdft_genus, XSIMD(dft_n1fsimd_genus)) = { n1f_okp, VL };
|
|
|
|
static int n2b_okp(const kdft_desc *d,
|
|
const R *ri, const R *ii, const R *ro, const R *io,
|
|
INT is, INT os, INT vl, INT ivs, INT ovs,
|
|
const planner *plnr)
|
|
{
|
|
return (1
|
|
&& ALIGNEDA(ii)
|
|
&& ALIGNEDA(io)
|
|
&& !NO_SIMDP(plnr)
|
|
&& SIMD_STRIDE_OKA(is)
|
|
&& SIMD_VSTRIDE_OKA(ivs)
|
|
&& SIMD_VSTRIDE_OKA(os) /* os == 2 enforced by codelet */
|
|
&& SIMD_STRIDE_OKPAIR(ovs)
|
|
&& ri == ii + 1
|
|
&& ro == io + 1
|
|
&& (vl % VL) == 0
|
|
&& (!d->is || (d->is == is))
|
|
&& (!d->os || (d->os == os))
|
|
&& (!d->ivs || (d->ivs == ivs))
|
|
&& (!d->ovs || (d->ovs == ovs))
|
|
);
|
|
}
|
|
|
|
EXTERN_CONST(kdft_genus, XSIMD(dft_n2bsimd_genus)) = { n2b_okp, VL };
|
|
|
|
static int n2f_okp(const kdft_desc *d,
|
|
const R *ri, const R *ii, const R *ro, const R *io,
|
|
INT is, INT os, INT vl, INT ivs, INT ovs,
|
|
const planner *plnr)
|
|
{
|
|
return (1
|
|
&& ALIGNEDA(ri)
|
|
&& ALIGNEDA(ro)
|
|
&& !NO_SIMDP(plnr)
|
|
&& SIMD_STRIDE_OKA(is)
|
|
&& SIMD_VSTRIDE_OKA(ivs)
|
|
&& SIMD_VSTRIDE_OKA(os) /* os == 2 enforced by codelet */
|
|
&& SIMD_STRIDE_OKPAIR(ovs)
|
|
&& ii == ri + 1
|
|
&& io == ro + 1
|
|
&& (vl % VL) == 0
|
|
&& (!d->is || (d->is == is))
|
|
&& (!d->os || (d->os == os))
|
|
&& (!d->ivs || (d->ivs == ivs))
|
|
&& (!d->ovs || (d->ovs == ovs))
|
|
);
|
|
}
|
|
|
|
EXTERN_CONST(kdft_genus, XSIMD(dft_n2fsimd_genus)) = { n2f_okp, VL };
|
|
|
|
static int n2s_okp(const kdft_desc *d,
|
|
const R *ri, const R *ii, const R *ro, const R *io,
|
|
INT is, INT os, INT vl, INT ivs, INT ovs,
|
|
const planner *plnr)
|
|
{
|
|
return (1
|
|
&& !NO_SIMDP(plnr)
|
|
&& ALIGNEDA(ri)
|
|
&& ALIGNEDA(ii)
|
|
&& ALIGNEDA(ro)
|
|
&& ALIGNEDA(io)
|
|
&& SIMD_STRIDE_OKA(is)
|
|
&& ivs == 1
|
|
&& os == 1
|
|
&& SIMD_STRIDE_OKA(ovs)
|
|
&& (vl % (2 * VL)) == 0
|
|
&& (!d->is || (d->is == is))
|
|
&& (!d->os || (d->os == os))
|
|
&& (!d->ivs || (d->ivs == ivs))
|
|
&& (!d->ovs || (d->ovs == ovs))
|
|
);
|
|
}
|
|
|
|
EXTERN_CONST(kdft_genus, XSIMD(dft_n2ssimd_genus)) = { n2s_okp, 2 * VL };
|
|
|
|
static int q1b_okp(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
return (1
|
|
&& ALIGNED(iio)
|
|
&& !NO_SIMDP(plnr)
|
|
&& SIMD_STRIDE_OK(rs)
|
|
&& SIMD_STRIDE_OK(vs)
|
|
&& SIMD_VSTRIDE_OK(ms)
|
|
&& rio == iio + 1
|
|
&& (m % VL) == 0
|
|
&& (mb % VL) == 0
|
|
&& (me % VL) == 0
|
|
&& (!d->rs || (d->rs == rs))
|
|
&& (!d->vs || (d->vs == vs))
|
|
&& (!d->ms || (d->ms == ms))
|
|
);
|
|
}
|
|
EXTERN_CONST(ct_genus, XSIMD(dft_q1bsimd_genus)) = { q1b_okp, VL };
|
|
|
|
static int q1f_okp(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
return (1
|
|
&& ALIGNED(rio)
|
|
&& !NO_SIMDP(plnr)
|
|
&& SIMD_STRIDE_OK(rs)
|
|
&& SIMD_STRIDE_OK(vs)
|
|
&& SIMD_VSTRIDE_OK(ms)
|
|
&& iio == rio + 1
|
|
&& (m % VL) == 0
|
|
&& (mb % VL) == 0
|
|
&& (me % VL) == 0
|
|
&& (!d->rs || (d->rs == rs))
|
|
&& (!d->vs || (d->vs == vs))
|
|
&& (!d->ms || (d->ms == ms))
|
|
);
|
|
}
|
|
EXTERN_CONST(ct_genus, XSIMD(dft_q1fsimd_genus)) = { q1f_okp, VL };
|
|
|
|
static int t_okp_common(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
UNUSED(rio); UNUSED(iio);
|
|
return (1
|
|
&& !NO_SIMDP(plnr)
|
|
&& SIMD_STRIDE_OKA(rs)
|
|
&& SIMD_VSTRIDE_OKA(ms)
|
|
&& (m % VL) == 0
|
|
&& (mb % VL) == 0
|
|
&& (me % VL) == 0
|
|
&& (!d->rs || (d->rs == rs))
|
|
&& (!d->vs || (d->vs == vs))
|
|
&& (!d->ms || (d->ms == ms))
|
|
);
|
|
}
|
|
|
|
static int t_okp_commonu(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
UNUSED(rio); UNUSED(iio); UNUSED(m);
|
|
return (1
|
|
&& !NO_SIMDP(plnr)
|
|
&& SIMD_STRIDE_OK(rs)
|
|
&& SIMD_VSTRIDE_OK(ms)
|
|
&& (mb % VL) == 0
|
|
&& (me % VL) == 0
|
|
&& (!d->rs || (d->rs == rs))
|
|
&& (!d->vs || (d->vs == vs))
|
|
&& (!d->ms || (d->ms == ms))
|
|
);
|
|
}
|
|
|
|
static int t_okp_t1f(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
return t_okp_common(d, rio, iio, rs, vs, m, mb, me, ms, plnr)
|
|
&& iio == rio + 1
|
|
&& ALIGNEDA(rio);
|
|
}
|
|
|
|
EXTERN_CONST(ct_genus, XSIMD(dft_t1fsimd_genus)) = { t_okp_t1f, VL };
|
|
|
|
static int t_okp_t1fu(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
return t_okp_commonu(d, rio, iio, rs, vs, m, mb, me, ms, plnr)
|
|
&& iio == rio + 1
|
|
&& ALIGNED(rio);
|
|
}
|
|
|
|
EXTERN_CONST(ct_genus, XSIMD(dft_t1fusimd_genus)) = { t_okp_t1fu, VL };
|
|
|
|
static int t_okp_t1b(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
return t_okp_common(d, rio, iio, rs, vs, m, mb, me, ms, plnr)
|
|
&& rio == iio + 1
|
|
&& ALIGNEDA(iio);
|
|
}
|
|
|
|
EXTERN_CONST(ct_genus, XSIMD(dft_t1bsimd_genus)) = { t_okp_t1b, VL };
|
|
|
|
static int t_okp_t1bu(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
return t_okp_commonu(d, rio, iio, rs, vs, m, mb, me, ms, plnr)
|
|
&& rio == iio + 1
|
|
&& ALIGNED(iio);
|
|
}
|
|
|
|
EXTERN_CONST(ct_genus, XSIMD(dft_t1busimd_genus)) = { t_okp_t1bu, VL };
|
|
|
|
/* use t2* codelets only when n = m*radix is small, because
|
|
t2* codelets use ~2n twiddle factors (instead of ~n) */
|
|
static int small_enough(const ct_desc *d, INT m)
|
|
{
|
|
return m * d->radix <= 16384;
|
|
}
|
|
|
|
static int t_okp_t2f(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
return t_okp_t1f(d, rio, iio, rs, vs, m, mb, me, ms, plnr)
|
|
&& small_enough(d, m);
|
|
}
|
|
|
|
EXTERN_CONST(ct_genus, XSIMD(dft_t2fsimd_genus)) = { t_okp_t2f, VL };
|
|
|
|
static int t_okp_t2b(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
return t_okp_t1b(d, rio, iio, rs, vs, m, mb, me, ms, plnr)
|
|
&& small_enough(d, m);
|
|
}
|
|
|
|
EXTERN_CONST(ct_genus, XSIMD(dft_t2bsimd_genus)) = { t_okp_t2b, VL };
|
|
|
|
static int ts_okp(const ct_desc *d,
|
|
const R *rio, const R *iio,
|
|
INT rs, INT vs, INT m, INT mb, INT me, INT ms,
|
|
const planner *plnr)
|
|
{
|
|
UNUSED(rio);
|
|
UNUSED(iio);
|
|
return (1
|
|
&& !NO_SIMDP(plnr)
|
|
&& ALIGNEDA(rio)
|
|
&& ALIGNEDA(iio)
|
|
&& SIMD_STRIDE_OKA(rs)
|
|
&& ms == 1
|
|
&& (m % (2 * VL)) == 0
|
|
&& (mb % (2 * VL)) == 0
|
|
&& (me % (2 * VL)) == 0
|
|
&& (!d->rs || (d->rs == rs))
|
|
&& (!d->vs || (d->vs == vs))
|
|
&& (!d->ms || (d->ms == ms))
|
|
);
|
|
}
|
|
|
|
EXTERN_CONST(ct_genus, XSIMD(dft_tssimd_genus)) = { ts_okp, 2 * VL };
|