/* * Copyright (c) 2003, 2007-11 Matteo Frigo * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology * * AVX-512 support implemented by Romain Dolbeau. * Romain Dolbeau hereby places his modifications in the public domain. * * Permission is hereby granted, free of charge, to any person obtaining a copy * of this software and associated documentation files (the "Software"), to deal * in the Software without restriction, including without limitation the rights * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell * copies of the Software, and to permit persons to whom the Software is * furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included in * all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN * THE SOFTWARE. * */ #if defined(FFTW_LDOUBLE) || defined(FFTW_QUAD) #error "AVX-512 vector instructions only works in single or double precision" #endif #ifdef FFTW_SINGLE # define DS(d,s) s /* single-precision option */ # define SUFF(name) name ## _ps # define SCAL(x) x ## f #else /* !FFTW_SINGLE */ # define DS(d,s) d /* double-precision option */ # define SUFF(name) name ## _pd # define SCAL(x) x #endif /* FFTW_SINGLE */ #define SIMD_SUFFIX _avx512 /* for renaming */ #define VL DS(4, 8) /* SIMD complex vector length */ #define SIMD_VSTRIDE_OKA(x) ((x) == 2) #define SIMD_STRIDE_OKPAIR SIMD_STRIDE_OK #if defined(__GNUC__) && !defined(__AVX512F__) /* sanity check */ #error "compiling simd-avx512.h without avx-512f support" #endif #if !defined(HAVE_AVX2) #warning "You should probably enable AVX2 with --enable-avx2 for AVX-512" #endif #include typedef DS(__m512d, __m512) V; #define VLIT(re, im) DS(SUFF(_mm512_setr)(im, re, im, re, im, re, im, re),SUFF(_mm512_setr)(im, re, im, re, im, re, im, re, im, re, im, re, im, re, im, re)) #define VLIT1(val) SUFF(_mm512_set1)(val) #define LDK(x) x #define DVK(var, val) V var = VLIT1(val) #define VZERO SUFF(_mm512_setzero)() #define VDUPL(x) DS(_mm512_movedup_pd(x),_mm512_moveldup_ps(x)) #define VDUPH(x) DS(_mm512_unpackhi_pd(x, x),_mm512_movehdup_ps(x)) #define FLIP_RI(x) SUFF(_mm512_shuffle)(x, x, DS(0x55,0xB1)) #define VCONJ(x) SUFF(_mm512_fmsubadd)(VZERO, VZERO, x) static inline V VBYI(V x) { return FLIP_RI(VCONJ(x)); } #define VADD(a,b) SUFF(_mm512_add)(a,b) #define VSUB(a,b) SUFF(_mm512_sub)(a,b) #define VMUL(a,b) SUFF(_mm512_mul)(a,b) #define VFMA(a, b, c) SUFF(_mm512_fmadd)(a, b, c) #define VFMS(a, b, c) SUFF(_mm512_fmsub)(a, b, c) #define VFNMS(a, b, c) SUFF(_mm512_fnmadd)(a, b, c) #define VFMAI(b, c) SUFF(_mm512_fmaddsub)(VLIT1(1.), c, FLIP_RI(b)) #define VFNMSI(b, c) SUFF(_mm512_fmsubadd)(VLIT1(1.), c, FLIP_RI(b)) #define VFMACONJ(b,c) SUFF(_mm512_fmsubadd)(VLIT1(1.), c, b) #define VFMSCONJ(b,c) SUFF(_mm512_fmsubadd)(VLIT1(-1.), c, b) #define VFNMSCONJ(b,c) SUFF(_mm512_fmaddsub)(VLIT1(1.), c, b) static inline V LDA(const R *x, INT ivs, const R *aligned_like) { (void)aligned_like; /* UNUSED */ (void)ivs; /* UNUSED */ return SUFF(_mm512_loadu)(x); } static inline void STA(R *x, V v, INT ovs, const R *aligned_like) { (void)aligned_like; /* UNUSED */ (void)ovs; /* UNUSED */ SUFF(_mm512_storeu)(x, v); } #if FFTW_SINGLE static inline V LDu(const R *x, INT ivs, const R *aligned_like) { (void)aligned_like; /* UNUSED */ __m512i index = _mm512_set_epi32(7 * ivs + 1, 7 * ivs, 6 * ivs + 1, 6 * ivs, 5 * ivs + 1, 5 * ivs, 4 * ivs + 1, 4 * ivs, 3 * ivs + 1, 3 * ivs, 2 * ivs + 1, 2 * ivs, 1 * ivs + 1, 1 * ivs, 0 * ivs + 1, 0 * ivs); return _mm512_i32gather_ps(index, x, 4); } static inline void STu(R *x, V v, INT ovs, const R *aligned_like) { (void)aligned_like; /* UNUSED */ __m512i index = _mm512_set_epi32(7 * ovs + 1, 7 * ovs, 6 * ovs + 1, 6 * ovs, 5 * ovs + 1, 5 * ovs, 4 * ovs + 1, 4 * ovs, 3 * ovs + 1, 3 * ovs, 2 * ovs + 1, 2 * ovs, 1 * ovs + 1, 1 * ovs, 0 * ovs + 1, 0 * ovs); _mm512_i32scatter_ps(x, index, v, 4); } #else /* !FFTW_SINGLE */ static inline V LDu(const R *x, INT ivs, const R *aligned_like) { (void)aligned_like; /* UNUSED */ __m256i index = _mm256_set_epi32(3 * ivs + 1, 3 * ivs, 2 * ivs + 1, 2 * ivs, 1 * ivs + 1, 1 * ivs, 0 * ivs + 1, 0 * ivs); return _mm512_i32gather_pd(index, x, 8); } static inline void STu(R *x, V v, INT ovs, const R *aligned_like) { (void)aligned_like; /* UNUSED */ __m256i index = _mm256_set_epi32(3 * ovs + 1, 3 * ovs, 2 * ovs + 1, 2 * ovs, 1 * ovs + 1, 1 * ovs, 0 * ovs + 1, 0 * ovs); _mm512_i32scatter_pd(x, index, v, 8); } #endif /* FFTW_SINGLE */ #define LD LDu #define ST STu #ifdef FFTW_SINGLE #define STM2(x, v, ovs, a) ST(x, v, ovs, a) #define STN2(x, v0, v1, ovs) /* nop */ static inline void STM4(R *x, V v, INT ovs, const R *aligned_like) { (void)aligned_like; /* UNUSED */ __m512i index = _mm512_set_epi32(15 * ovs, 14 * ovs, 13 * ovs, 12 * ovs, 11 * ovs, 10 * ovs, 9 * ovs, 8 * ovs, 7 * ovs, 6 * ovs, 5 * ovs, 4 * ovs, 3 * ovs, 2 * ovs, 1 * ovs, 0 * ovs); _mm512_i32scatter_ps(x, index, v, 4); } #define STN4(x, v0, v1, v2, v3, ovs) /* no-op */ #else /* !FFTW_SINGLE */ #define STM2(x, v, ovs, a) ST(x, v, ovs, a) #define STN2(x, v0, v1, ovs) /* nop */ static inline void STM4(R *x, V v, INT ovs, const R *aligned_like) { (void)aligned_like; /* UNUSED */ __m256i index = _mm256_set_epi32(7 * ovs, 6 * ovs, 5 * ovs, 4 * ovs, 3 * ovs, 2 * ovs, 1 * ovs, 0 * ovs); _mm512_i32scatter_pd(x, index, v, 8); } #define STN4(x, v0, v1, v2, v3, ovs) /* no-op */ #endif /* FFTW_SINGLE */ static inline V VZMUL(V tx, V sr) { /* V tr = VDUPL(tx); */ /* V ti = VDUPH(tx); */ /* tr = VMUL(sr, tr); */ /* sr = VBYI(sr); */ /* return VFMA(ti, sr, tr); */ return SUFF(_mm512_fmaddsub)(sr, VDUPL(tx), VMUL(FLIP_RI(sr), VDUPH(tx))); } static inline V VZMULJ(V tx, V sr) { /* V tr = VDUPL(tx); */ /* V ti = VDUPH(tx); */ /* tr = VMUL(sr, tr); */ /* sr = VBYI(sr); */ /* return VFNMS(ti, sr, tr); */ return SUFF(_mm512_fmsubadd)(sr, VDUPL(tx), VMUL(FLIP_RI(sr), VDUPH(tx))); } static inline V VZMULI(V tx, V sr) { V tr = VDUPL(tx); V ti = VDUPH(tx); ti = VMUL(ti, sr); sr = VBYI(sr); return VFMS(tr, sr, ti); /* return SUFF(_mm512_addsub)(SUFF(_mm512_fnmadd)(sr, VDUPH(tx), VZERO), VMUL(FLIP_RI(sr), VDUPL(tx))); */ } static inline V VZMULIJ(V tx, V sr) { /* V tr = VDUPL(tx); */ /* V ti = VDUPH(tx); */ /* ti = VMUL(ti, sr); */ /* sr = VBYI(sr); */ /* return VFMA(tr, sr, ti); */ return SUFF(_mm512_fmaddsub)(sr, VDUPH(tx), VMUL(FLIP_RI(sr), VDUPL(tx))); } /* twiddle storage #1: compact, slower */ #ifdef FFTW_SINGLE # define VTW1(v,x) {TW_CEXP, v, x}, {TW_CEXP, v+1, x}, {TW_CEXP, v+2, x}, {TW_CEXP, v+3, x}, {TW_CEXP, v+4, x}, {TW_CEXP, v+5, x}, {TW_CEXP, v+6, x}, {TW_CEXP, v+7, x} #else /* !FFTW_SINGLE */ # define VTW1(v,x) {TW_CEXP, v, x}, {TW_CEXP, v+1, x}, {TW_CEXP, v+2, x}, {TW_CEXP, v+3, x} #endif /* FFTW_SINGLE */ #define TWVL1 (VL) static inline V BYTW1(const R *t, V sr) { return VZMUL(LDA(t, 2, t), sr); } static inline V BYTWJ1(const R *t, V sr) { return VZMULJ(LDA(t, 2, t), sr); } /* twiddle storage #2: twice the space, faster (when in cache) */ #ifdef FFTW_SINGLE # define VTW2(v,x) \ {TW_COS, v , x}, {TW_COS, v , x}, {TW_COS, v+1, x}, {TW_COS, v+1, x}, \ {TW_COS, v+2, x}, {TW_COS, v+2, x}, {TW_COS, v+3, x}, {TW_COS, v+3, x}, \ {TW_COS, v+4, x}, {TW_COS, v+4, x}, {TW_COS, v+5, x}, {TW_COS, v+5, x}, \ {TW_COS, v+6, x}, {TW_COS, v+6, x}, {TW_COS, v+7, x}, {TW_COS, v+7, x}, \ {TW_SIN, v , -x}, {TW_SIN, v , x}, {TW_SIN, v+1, -x}, {TW_SIN, v+1, x}, \ {TW_SIN, v+2, -x}, {TW_SIN, v+2, x}, {TW_SIN, v+3, -x}, {TW_SIN, v+3, x}, \ {TW_SIN, v+4, -x}, {TW_SIN, v+4, x}, {TW_SIN, v+5, -x}, {TW_SIN, v+5, x}, \ {TW_SIN, v+6, -x}, {TW_SIN, v+6, x}, {TW_SIN, v+7, -x}, {TW_SIN, v+7, x} #else /* !FFTW_SINGLE */ # define VTW2(v,x) \ {TW_COS, v , x}, {TW_COS, v , x}, {TW_COS, v+1, x}, {TW_COS, v+1, x}, \ {TW_COS, v+2, x}, {TW_COS, v+2, x}, {TW_COS, v+3, x}, {TW_COS, v+3, x}, \ {TW_SIN, v , -x}, {TW_SIN, v , x}, {TW_SIN, v+1, -x}, {TW_SIN, v+1, x}, \ {TW_SIN, v+2, -x}, {TW_SIN, v+2, x}, {TW_SIN, v+3, -x}, {TW_SIN, v+3, x} #endif /* FFTW_SINGLE */ #define TWVL2 (2 * VL) static inline V BYTW2(const R *t, V sr) { const V *twp = (const V *)t; V si = FLIP_RI(sr); V tr = twp[0], ti = twp[1]; /* V tr = LD(t, 2, t), ti = LD(t + VL, 2, t + VL); */ return VFMA(tr, sr, VMUL(ti, si)); } static inline V BYTWJ2(const R *t, V sr) { const V *twp = (const V *)t; V si = FLIP_RI(sr); V tr = twp[0], ti = twp[1]; /* V tr = LD(t, 2, t), ti = LD(t + VL, 2, t + VL); */ return VFNMS(ti, si, VMUL(tr, sr)); } /* twiddle storage #3 */ #define VTW3(v,x) VTW1(v,x) #define TWVL3 TWVL1 /* twiddle storage for split arrays */ #ifdef FFTW_SINGLE # define VTWS(v,x) \ {TW_COS, v , x}, {TW_COS, v+1 , x}, {TW_COS, v+2 , x}, {TW_COS, v+3 , x}, \ {TW_COS, v+4 , x}, {TW_COS, v+5 , x}, {TW_COS, v+6 , x}, {TW_COS, v+7 , x}, \ {TW_COS, v+8 , x}, {TW_COS, v+9 , x}, {TW_COS, v+10, x}, {TW_COS, v+11, x}, \ {TW_COS, v+12, x}, {TW_COS, v+13, x}, {TW_COS, v+14, x}, {TW_COS, v+15, x}, \ {TW_SIN, v , x}, {TW_SIN, v+1 , x}, {TW_SIN, v+2 , x}, {TW_SIN, v+3 , x}, \ {TW_SIN, v+4 , x}, {TW_SIN, v+5 , x}, {TW_SIN, v+6 , x}, {TW_SIN, v+7 , x}, \ {TW_SIN, v+8 , x}, {TW_SIN, v+9 , x}, {TW_SIN, v+10, x}, {TW_SIN, v+11, x}, \ {TW_SIN, v+12, x}, {TW_SIN, v+13, x}, {TW_SIN, v+14, x}, {TW_SIN, v+15, x} #else /* !FFTW_SINGLE */ # define VTWS(v,x) \ {TW_COS, v , x}, {TW_COS, v+1, x}, {TW_COS, v+2, x}, {TW_COS, v+3, x}, \ {TW_COS, v+4, x}, {TW_COS, v+5, x}, {TW_COS, v+6, x}, {TW_COS, v+7, x}, \ {TW_SIN, v , x}, {TW_SIN, v+1, x}, {TW_SIN, v+2, x}, {TW_SIN, v+3, x}, \ {TW_SIN, v+4, x}, {TW_SIN, v+5, x}, {TW_SIN, v+6, x}, {TW_SIN, v+7, x} #endif /* FFTW_SINGLE */ #define TWVLS (2 * VL) #define VLEAVE _mm256_zeroupper #include "simd-common.h"