include/NTL/PD.h

11 struct PD {  struct
13    PD();
30 struct PD<8> {
35    PD() { }
36    PD(double x) : data(_mm512_set1_pd(x)) { }
37    PD(__m512d _data) : data(_data) { }
39    PD(double d0, double d1, double d2, double d3,
43    static PD load(const double *p) { return _mm512_load_pd(p); }
46    static PD loadu(const double *p) { return _mm512_loadu_pd(p); }
50 load(PD<8>& x, const double *p)
51 { x = PD<8>::load(p); }
55 loadu(PD<8>& x, const double *p)
56 { x = PD<8>::loadu(p); }
59 store(double *p, PD<8> a)
64 storeu(double *p, PD<8> a)
69 load(PD<8>& x, const long *p)
74 loadu(PD<8>& x, const long *p)
79 store(long *p, PD<8> a)
84 storeu(long *p, PD<8> a)
90 inline PD<8>
91 swap2(PD<8> a)
96 inline PD<8>
97 swap4(PD<8> a)
101 inline PD<8>
102 dup2even(PD<8> a)
106 inline PD<8>
107 dup2odd(PD<8> a)
111 inline PD<8>
112 dup4even(PD<8> a)
116 inline PD<8>
117 dup4odd(PD<8> a)
122 inline PD<8>
123 blend2(PD<8> a, PD<8> b)
129 inline PD<8>
130 blend4(PD<8> a, PD<8> b)
135 inline PD<8>
136 correct_excess(PD<8> a, PD<8> b)
143 inline PD<8>
144 correct_deficit(PD<8> a, PD<8> b)
151 clear(PD<8>& x)
154 inline PD<8>
155 operator+(PD<8> a, PD<8> b)
158 inline PD<8>
159 operator-(PD<8> a, PD<8> b)
162 inline PD<8>
163 operator*(PD<8> a, PD<8> b)
166 inline PD<8>
167 operator/(PD<8> a, PD<8> b)
170 inline PD<8>&
171 operator+=(PD<8>& a, PD<8> b)
174 inline PD<8>&
175 operator-=(PD<8>& a, PD<8> b)
178 inline PD<8>&
179 operator*=(PD<8>& a, PD<8> b)
182 inline PD<8>&
183 operator/=(PD<8>& a, PD<8> b)
187 inline PD<8>
188 fused_muladd(PD<8> a, PD<8> b, PD<8> c)
192 inline PD<8>
193 fused_mulsub(PD<8> a, PD<8> b, PD<8> c)
197 inline PD<8>
198 fused_negmuladd(PD<8> a, PD<8> b, PD<8> c)
208 struct PD<4> {
213    PD() { }
214    PD(double x) : data(_mm256_set1_pd(x)) { }
215    PD(__m256d _data) : data(_data) { }
216    PD(double d0, double d1, double d2, double d3)
219    static PD load(const double *p) { return _mm256_load_pd(p); }
222    static PD loadu(const double *p) { return _mm256_loadu_pd(p); }
226 load(PD<4>& x, const double *p)
227 { x = PD<4>::load(p); }
231 loadu(PD<4>& x, const double *p)
232 { x = PD<4>::loadu(p); }
235 store(double *p, PD<4> a)
240 storeu(double *p, PD<4> a)
262 load(PD<4>& x, const long *p)
276 loadu(PD<4>& x, const long *p)
289 store(long *p, PD<4> a)
311 storeu(long *p, PD<4> a)
329 inline PD<4>
330 swap2(PD<4> a)
334 inline PD<4>
335 dup2even(PD<4> a)
339 inline PD<4>
340 dup2odd(PD<4> a)
345 inline PD<4>
346 blend2(PD<4> a, PD<4> b)
350 inline PD<4>
351 correct_excess(PD<4> a, PD<4> b)
364 inline PD<4>
365 correct_deficit(PD<4> a, PD<4> b)
378 clear(PD<4>& x)
381 inline PD<4>
382 operator+(PD<4> a, PD<4> b)
385 inline PD<4>
386 operator-(PD<4> a, PD<4> b)
389 inline PD<4>
390 operator*(PD<4> a, PD<4> b)
393 inline PD<4>
394 operator/(PD<4> a, PD<4> b)
397 inline PD<4>&
398 operator+=(PD<4>& a, PD<4> b)
401 inline PD<4>&
402 operator-=(PD<4>& a, PD<4> b)
405 inline PD<4>&
406 operator*=(PD<4>& a, PD<4> b)
409 inline PD<4>&
410 operator/=(PD<4>& a, PD<4> b)
414 inline PD<4>
415 fused_muladd(PD<4> a, PD<4> b, PD<4> c)
419 inline PD<4>
420 fused_mulsub(PD<4> a, PD<4> b, PD<4> c)
424 inline PD<4>
425 fused_negmuladd(PD<4> a, PD<4> b, PD<4> c)
433 struct PD<2> {
438    PD() { }
439    PD(double x) : data(_mm_set1_pd(x)) { }
440    PD(__m128d _data) : data(_data) { }
441    PD(double d0, double d1)
444    static PD load(const double *p) { return _mm_load_pd(p); }
447    static PD loadu(const double *p) { return _mm_loadu_pd(p); }
451 load(PD<2>& x, const double *p)
452 { x = PD<2>::load(p); }
456 loadu(PD<2>& x, const double *p)
457 { x = PD<2>::loadu(p); }
460 store(double *p, PD<2> a)
465 storeu(double *p, PD<2> a)
478 load(PD<2>& x, const long *p)
492 loadu(PD<2>& x, const long *p)
505 store(long *p, PD<2> a)
527 storeu(long *p, PD<2> a)
544 inline PD<2>
545 correct_excess(PD<2> a, PD<2> b)
558 inline PD<2>
559 correct_deficit(PD<2> a, PD<2> b)
572 clear(PD<2>& x)
575 inline PD<2>
576 operator+(PD<2> a, PD<2> b)
579 inline PD<2>
580 operator-(PD<2> a, PD<2> b)
583 inline PD<2>
584 operator*(PD<2> a, PD<2> b)
587 inline PD<2>
588 operator/(PD<2> a, PD<2> b)
591 inline PD<2>&
592 operator+=(PD<2>& a, PD<2> b)
595 inline PD<2>&
596 operator-=(PD<2>& a, PD<2> b)
599 inline PD<2>&
600 operator*=(PD<2>& a, PD<2> b)
603 inline PD<2>&
604 operator/=(PD<2>& a, PD<2> b)
608 inline PD<2>
609 fused_muladd(PD<2> a, PD<2> b, PD<2> c)
613 inline PD<2>
614 fused_mulsub(PD<2> a, PD<2> b, PD<2> c)
618 inline PD<2>
619 fused_negmuladd(PD<2> a, PD<2> b, PD<2> c)
630 inline PD<8>
631 join(PD<4> a, PD<4> b)
638 inline PD<4>
639 get_lo(PD<8> a)
643 inline PD<4>
644 get_hi(PD<8> a)
652 inline PD<4>
653 join(PD<2> a, PD<2> b)
663 inline PD<2>
664 get_lo(PD<4> a)
668 inline PD<2>
669 get_hi(PD<4> a)