1// This file is part of Eigen, a lightweight C++ template library
2// for linear algebra.
3//
4// Copyright (C) 2016 Benoit Steiner <benoit.steiner.goog@gmail.com>
5//
6// This Source Code Form is subject to the terms of the Mozilla
7// Public License v. 2.0. If a copy of the MPL was not distributed
8// with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
9
10#ifndef EIGEN_PACKET_MATH_HALF_CUDA_H
11#define EIGEN_PACKET_MATH_HALF_CUDA_H
12
13
14namespace Eigen {
15namespace internal {
16
17// Most of the following operations require arch >= 3.0
18#if defined(EIGEN_HAS_CUDA_FP16) && defined(__CUDACC__) && defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 300
19
20template<> struct is_arithmetic<half2> { enum { value = true }; };
21
22template<> struct packet_traits<Eigen::half> : default_packet_traits
23{
24 typedef half2 type;
25 typedef half2 half;
26 enum {
27 Vectorizable = 1,
28 AlignedOnScalar = 1,
29 size=2,
30 HasHalfPacket = 0,
31 HasAdd = 1,
32 HasMul = 1,
33 HasDiv = 1,
34 HasSqrt = 1,
35 HasRsqrt = 1,
36 HasExp = 1,
37 HasLog = 1,
38 HasLog1p = 1
39 };
40};
41
42template<> struct unpacket_traits<half2> { typedef Eigen::half type; enum {size=2, alignment=Aligned16}; typedef half2 half; };
43
44template<> __device__ EIGEN_STRONG_INLINE half2 pset1<half2>(const Eigen::half& from) {
45 return __half2half2(from);
46}
47
48template<> __device__ EIGEN_STRONG_INLINE half2 pload<half2>(const Eigen::half* from) {
49 return *reinterpret_cast<const half2*>(from);
50}
51
52template<> __device__ EIGEN_STRONG_INLINE half2 ploadu<half2>(const Eigen::half* from) {
53 return __halves2half2(from[0], from[1]);
54}
55
56template<> EIGEN_STRONG_INLINE half2 ploaddup<half2>(const Eigen::half* from) {
57 return __halves2half2(from[0], from[0]);
58}
59
60template<> __device__ EIGEN_STRONG_INLINE void pstore<Eigen::half>(Eigen::half* to, const half2& from) {
61 *reinterpret_cast<half2*>(to) = from;
62}
63
64template<> __device__ EIGEN_STRONG_INLINE void pstoreu<Eigen::half>(Eigen::half* to, const half2& from) {
65 to[0] = __low2half(from);
66 to[1] = __high2half(from);
67}
68
69template<>
70 __device__ EIGEN_ALWAYS_INLINE half2 ploadt_ro<half2, Aligned>(const Eigen::half* from) {
71#if __CUDA_ARCH__ >= 350
72 return __ldg((const half2*)from);
73#else
74 return __halves2half2(*(from+0), *(from+1));
75#endif
76}
77
78template<>
79__device__ EIGEN_ALWAYS_INLINE half2 ploadt_ro<half2, Unaligned>(const Eigen::half* from) {
80#if __CUDA_ARCH__ >= 350
81 return __halves2half2(__ldg(from+0), __ldg(from+1));
82#else
83 return __halves2half2(*(from+0), *(from+1));
84#endif
85}
86
87template<> __device__ EIGEN_STRONG_INLINE half2 pgather<Eigen::half, half2>(const Eigen::half* from, Index stride) {
88 return __halves2half2(from[0*stride], from[1*stride]);
89}
90
91template<> __device__ EIGEN_STRONG_INLINE void pscatter<Eigen::half, half2>(Eigen::half* to, const half2& from, Index stride) {
92 to[stride*0] = __low2half(from);
93 to[stride*1] = __high2half(from);
94}
95
96template<> __device__ EIGEN_STRONG_INLINE Eigen::half pfirst<half2>(const half2& a) {
97 return __low2half(a);
98}
99
100template<> __device__ EIGEN_STRONG_INLINE half2 pabs<half2>(const half2& a) {
101 half2 result;
102 unsigned temp = *(reinterpret_cast<const unsigned*>(&(a)));
103 *(reinterpret_cast<unsigned*>(&(result))) = temp & 0x7FFF7FFF;
104 return result;
105}
106
107
108__device__ EIGEN_STRONG_INLINE void
109ptranspose(PacketBlock<half2,2>& kernel) {
110 __half a1 = __low2half(kernel.packet[0]);
111 __half a2 = __high2half(kernel.packet[0]);
112 __half b1 = __low2half(kernel.packet[1]);
113 __half b2 = __high2half(kernel.packet[1]);
114 kernel.packet[0] = __halves2half2(a1, b1);
115 kernel.packet[1] = __halves2half2(a2, b2);
116}
117
118template<> __device__ EIGEN_STRONG_INLINE half2 plset<half2>(const Eigen::half& a) {
119#if __CUDA_ARCH__ >= 530
120 return __halves2half2(a, __hadd(a, __float2half(1.0f)));
121#else
122 float f = __half2float(a) + 1.0f;
123 return __halves2half2(a, __float2half(f));
124#endif
125}
126
127template<> __device__ EIGEN_STRONG_INLINE half2 padd<half2>(const half2& a, const half2& b) {
128#if __CUDA_ARCH__ >= 530
129 return __hadd2(a, b);
130#else
131 float a1 = __low2float(a);
132 float a2 = __high2float(a);
133 float b1 = __low2float(b);
134 float b2 = __high2float(b);
135 float r1 = a1 + b1;
136 float r2 = a2 + b2;
137 return __floats2half2_rn(r1, r2);
138#endif
139}
140
141template<> __device__ EIGEN_STRONG_INLINE half2 psub<half2>(const half2& a, const half2& b) {
142#if __CUDA_ARCH__ >= 530
143 return __hsub2(a, b);
144#else
145 float a1 = __low2float(a);
146 float a2 = __high2float(a);
147 float b1 = __low2float(b);
148 float b2 = __high2float(b);
149 float r1 = a1 - b1;
150 float r2 = a2 - b2;
151 return __floats2half2_rn(r1, r2);
152#endif
153}
154
155template<> __device__ EIGEN_STRONG_INLINE half2 pnegate(const half2& a) {
156#if __CUDA_ARCH__ >= 530
157 return __hneg2(a);
158#else
159 float a1 = __low2float(a);
160 float a2 = __high2float(a);
161 return __floats2half2_rn(-a1, -a2);
162#endif
163}
164
165template<> __device__ EIGEN_STRONG_INLINE half2 pconj(const half2& a) { return a; }
166
167template<> __device__ EIGEN_STRONG_INLINE half2 pmul<half2>(const half2& a, const half2& b) {
168#if __CUDA_ARCH__ >= 530
169 return __hmul2(a, b);
170#else
171 float a1 = __low2float(a);
172 float a2 = __high2float(a);
173 float b1 = __low2float(b);
174 float b2 = __high2float(b);
175 float r1 = a1 * b1;
176 float r2 = a2 * b2;
177 return __floats2half2_rn(r1, r2);
178#endif
179}
180
181template<> __device__ EIGEN_STRONG_INLINE half2 pmadd<half2>(const half2& a, const half2& b, const half2& c) {
182#if __CUDA_ARCH__ >= 530
183 return __hfma2(a, b, c);
184#else
185 float a1 = __low2float(a);
186 float a2 = __high2float(a);
187 float b1 = __low2float(b);
188 float b2 = __high2float(b);
189 float c1 = __low2float(c);
190 float c2 = __high2float(c);
191 float r1 = a1 * b1 + c1;
192 float r2 = a2 * b2 + c2;
193 return __floats2half2_rn(r1, r2);
194#endif
195}
196
197template<> __device__ EIGEN_STRONG_INLINE half2 pdiv<half2>(const half2& a, const half2& b) {
198 float a1 = __low2float(a);
199 float a2 = __high2float(a);
200 float b1 = __low2float(b);
201 float b2 = __high2float(b);
202 float r1 = a1 / b1;
203 float r2 = a2 / b2;
204 return __floats2half2_rn(r1, r2);
205}
206
207template<> __device__ EIGEN_STRONG_INLINE half2 pmin<half2>(const half2& a, const half2& b) {
208 float a1 = __low2float(a);
209 float a2 = __high2float(a);
210 float b1 = __low2float(b);
211 float b2 = __high2float(b);
212 __half r1 = a1 < b1 ? __low2half(a) : __low2half(b);
213 __half r2 = a2 < b2 ? __high2half(a) : __high2half(b);
214 return __halves2half2(r1, r2);
215}
216
217template<> __device__ EIGEN_STRONG_INLINE half2 pmax<half2>(const half2& a, const half2& b) {
218 float a1 = __low2float(a);
219 float a2 = __high2float(a);
220 float b1 = __low2float(b);
221 float b2 = __high2float(b);
222 __half r1 = a1 > b1 ? __low2half(a) : __low2half(b);
223 __half r2 = a2 > b2 ? __high2half(a) : __high2half(b);
224 return __halves2half2(r1, r2);
225}
226
227template<> __device__ EIGEN_STRONG_INLINE Eigen::half predux<half2>(const half2& a) {
228#if __CUDA_ARCH__ >= 530
229 return __hadd(__low2half(a), __high2half(a));
230#else
231 float a1 = __low2float(a);
232 float a2 = __high2float(a);
233 return Eigen::half(half_impl::raw_uint16_to_half(__float2half_rn(a1 + a2)));
234#endif
235}
236
237template<> __device__ EIGEN_STRONG_INLINE Eigen::half predux_max<half2>(const half2& a) {
238#if __CUDA_ARCH__ >= 530
239 __half first = __low2half(a);
240 __half second = __high2half(a);
241 return __hgt(first, second) ? first : second;
242#else
243 float a1 = __low2float(a);
244 float a2 = __high2float(a);
245 return a1 > a2 ? __low2half(a) : __high2half(a);
246#endif
247}
248
249template<> __device__ EIGEN_STRONG_INLINE Eigen::half predux_min<half2>(const half2& a) {
250#if __CUDA_ARCH__ >= 530
251 __half first = __low2half(a);
252 __half second = __high2half(a);
253 return __hlt(first, second) ? first : second;
254#else
255 float a1 = __low2float(a);
256 float a2 = __high2float(a);
257 return a1 < a2 ? __low2half(a) : __high2half(a);
258#endif
259}
260
261template<> __device__ EIGEN_STRONG_INLINE Eigen::half predux_mul<half2>(const half2& a) {
262#if __CUDA_ARCH__ >= 530
263 return __hmul(__low2half(a), __high2half(a));
264#else
265 float a1 = __low2float(a);
266 float a2 = __high2float(a);
267 return Eigen::half(half_impl::raw_uint16_to_half(__float2half_rn(a1 * a2)));
268#endif
269}
270
271template<> __device__ EIGEN_STRONG_INLINE half2 plog1p<half2>(const half2& a) {
272 float a1 = __low2float(a);
273 float a2 = __high2float(a);
274 float r1 = log1pf(a1);
275 float r2 = log1pf(a2);
276 return __floats2half2_rn(r1, r2);
277}
278
279#if EIGEN_CUDACC_VER >= 80000 && defined EIGEN_CUDA_ARCH && EIGEN_CUDA_ARCH >= 530
280
281template<> __device__ EIGEN_STRONG_INLINE
282half2 plog<half2>(const half2& a) {
283 return h2log(a);
284}
285
286template<> __device__ EIGEN_STRONG_INLINE
287half2 pexp<half2>(const half2& a) {
288 return h2exp(a);
289}
290
291template<> __device__ EIGEN_STRONG_INLINE
292half2 psqrt<half2>(const half2& a) {
293 return h2sqrt(a);
294}
295
296template<> __device__ EIGEN_STRONG_INLINE
297half2 prsqrt<half2>(const half2& a) {
298 return h2rsqrt(a);
299}
300
301#else
302
303template<> __device__ EIGEN_STRONG_INLINE half2 plog<half2>(const half2& a) {
304 float a1 = __low2float(a);
305 float a2 = __high2float(a);
306 float r1 = logf(a1);
307 float r2 = logf(a2);
308 return __floats2half2_rn(r1, r2);
309}
310
311template<> __device__ EIGEN_STRONG_INLINE half2 pexp<half2>(const half2& a) {
312 float a1 = __low2float(a);
313 float a2 = __high2float(a);
314 float r1 = expf(a1);
315 float r2 = expf(a2);
316 return __floats2half2_rn(r1, r2);
317}
318
319template<> __device__ EIGEN_STRONG_INLINE half2 psqrt<half2>(const half2& a) {
320 float a1 = __low2float(a);
321 float a2 = __high2float(a);
322 float r1 = sqrtf(a1);
323 float r2 = sqrtf(a2);
324 return __floats2half2_rn(r1, r2);
325}
326
327template<> __device__ EIGEN_STRONG_INLINE half2 prsqrt<half2>(const half2& a) {
328 float a1 = __low2float(a);
329 float a2 = __high2float(a);
330 float r1 = rsqrtf(a1);
331 float r2 = rsqrtf(a2);
332 return __floats2half2_rn(r1, r2);
333}
334
335#endif
336
337#elif defined EIGEN_VECTORIZE_AVX512
338
339typedef struct {
340 __m256i x;
341} Packet16h;
342
343
344template<> struct is_arithmetic<Packet16h> { enum { value = true }; };
345
346template <>
347struct packet_traits<half> : default_packet_traits {
348 typedef Packet16h type;
349 // There is no half-size packet for Packet16h.
350 typedef Packet16h half;
351 enum {
352 Vectorizable = 1,
353 AlignedOnScalar = 1,
354 size = 16,
355 HasHalfPacket = 0,
356 HasAdd = 0,
357 HasSub = 0,
358 HasMul = 0,
359 HasNegate = 0,
360 HasAbs = 0,
361 HasAbs2 = 0,
362 HasMin = 0,
363 HasMax = 0,
364 HasConj = 0,
365 HasSetLinear = 0,
366 HasDiv = 0,
367 HasSqrt = 0,
368 HasRsqrt = 0,
369 HasExp = 0,
370 HasLog = 0,
371 HasBlend = 0
372 };
373};
374
375
376template<> struct unpacket_traits<Packet16h> { typedef Eigen::half type; enum {size=16, alignment=Aligned32}; typedef Packet16h half; };
377
378template<> EIGEN_STRONG_INLINE Packet16h pset1<Packet16h>(const Eigen::half& from) {
379 Packet16h result;
380 result.x = _mm256_set1_epi16(from.x);
381 return result;
382}
383
384template<> EIGEN_STRONG_INLINE Eigen::half pfirst<Packet16h>(const Packet16h& from) {
385 return half_impl::raw_uint16_to_half(static_cast<unsigned short>(_mm256_extract_epi16(from.x, 0)));
386}
387
388template<> EIGEN_STRONG_INLINE Packet16h pload<Packet16h>(const Eigen::half* from) {
389 Packet16h result;
390 result.x = _mm256_load_si256(reinterpret_cast<const __m256i*>(from));
391 return result;
392}
393
394template<> EIGEN_STRONG_INLINE Packet16h ploadu<Packet16h>(const Eigen::half* from) {
395 Packet16h result;
396 result.x = _mm256_loadu_si256(reinterpret_cast<const __m256i*>(from));
397 return result;
398}
399
400template<> EIGEN_STRONG_INLINE void pstore<half>(Eigen::half* to, const Packet16h& from) {
401 _mm256_store_si256((__m256i*)to, from.x);
402}
403
404template<> EIGEN_STRONG_INLINE void pstoreu<half>(Eigen::half* to, const Packet16h& from) {
405 _mm256_storeu_si256((__m256i*)to, from.x);
406}
407
408template<> EIGEN_STRONG_INLINE Packet16h
409ploadquad(const Eigen::half* from) {
410 Packet16h result;
411 unsigned short a = from[0].x;
412 unsigned short b = from[1].x;
413 unsigned short c = from[2].x;
414 unsigned short d = from[3].x;
415 result.x = _mm256_set_epi16(d, d, d, d, c, c, c, c, b, b, b, b, a, a, a, a);
416 return result;
417}
418
419EIGEN_STRONG_INLINE Packet16f half2float(const Packet16h& a) {
420#ifdef EIGEN_HAS_FP16_C
421 return _mm512_cvtph_ps(a.x);
422#else
423 EIGEN_ALIGN64 half aux[16];
424 pstore(aux, a);
425 float f0(aux[0]);
426 float f1(aux[1]);
427 float f2(aux[2]);
428 float f3(aux[3]);
429 float f4(aux[4]);
430 float f5(aux[5]);
431 float f6(aux[6]);
432 float f7(aux[7]);
433 float f8(aux[8]);
434 float f9(aux[9]);
435 float fa(aux[10]);
436 float fb(aux[11]);
437 float fc(aux[12]);
438 float fd(aux[13]);
439 float fe(aux[14]);
440 float ff(aux[15]);
441
442 return _mm512_set_ps(
443 ff, fe, fd, fc, fb, fa, f9, f8, f7, f6, f5, f4, f3, f2, f1, f0);
444#endif
445}
446
447EIGEN_STRONG_INLINE Packet16h float2half(const Packet16f& a) {
448#ifdef EIGEN_HAS_FP16_C
449 Packet16h result;
450 result.x = _mm512_cvtps_ph(a, _MM_FROUND_TO_NEAREST_INT|_MM_FROUND_NO_EXC);
451 return result;
452#else
453 EIGEN_ALIGN64 float aux[16];
454 pstore(aux, a);
455 half h0(aux[0]);
456 half h1(aux[1]);
457 half h2(aux[2]);
458 half h3(aux[3]);
459 half h4(aux[4]);
460 half h5(aux[5]);
461 half h6(aux[6]);
462 half h7(aux[7]);
463 half h8(aux[8]);
464 half h9(aux[9]);
465 half ha(aux[10]);
466 half hb(aux[11]);
467 half hc(aux[12]);
468 half hd(aux[13]);
469 half he(aux[14]);
470 half hf(aux[15]);
471
472 Packet16h result;
473 result.x = _mm256_set_epi16(
474 hf.x, he.x, hd.x, hc.x, hb.x, ha.x, h9.x, h8.x,
475 h7.x, h6.x, h5.x, h4.x, h3.x, h2.x, h1.x, h0.x);
476 return result;
477#endif
478}
479
480template<> EIGEN_STRONG_INLINE Packet16h padd<Packet16h>(const Packet16h& a, const Packet16h& b) {
481 Packet16f af = half2float(a);
482 Packet16f bf = half2float(b);
483 Packet16f rf = padd(af, bf);
484 return float2half(rf);
485}
486
487template<> EIGEN_STRONG_INLINE Packet16h pmul<Packet16h>(const Packet16h& a, const Packet16h& b) {
488 Packet16f af = half2float(a);
489 Packet16f bf = half2float(b);
490 Packet16f rf = pmul(af, bf);
491 return float2half(rf);
492}
493
494template<> EIGEN_STRONG_INLINE half predux<Packet16h>(const Packet16h& from) {
495 Packet16f from_float = half2float(from);
496 return half(predux(from_float));
497}
498
499template<> EIGEN_STRONG_INLINE Packet16h pgather<Eigen::half, Packet16h>(const Eigen::half* from, Index stride)
500{
501 Packet16h result;
502 result.x = _mm256_set_epi16(
503 from[15*stride].x, from[14*stride].x, from[13*stride].x, from[12*stride].x,
504 from[11*stride].x, from[10*stride].x, from[9*stride].x, from[8*stride].x,
505 from[7*stride].x, from[6*stride].x, from[5*stride].x, from[4*stride].x,
506 from[3*stride].x, from[2*stride].x, from[1*stride].x, from[0*stride].x);
507 return result;
508}
509
510template<> EIGEN_STRONG_INLINE void pscatter<half, Packet16h>(half* to, const Packet16h& from, Index stride)
511{
512 EIGEN_ALIGN64 half aux[16];
513 pstore(aux, from);
514 to[stride*0].x = aux[0].x;
515 to[stride*1].x = aux[1].x;
516 to[stride*2].x = aux[2].x;
517 to[stride*3].x = aux[3].x;
518 to[stride*4].x = aux[4].x;
519 to[stride*5].x = aux[5].x;
520 to[stride*6].x = aux[6].x;
521 to[stride*7].x = aux[7].x;
522 to[stride*8].x = aux[8].x;
523 to[stride*9].x = aux[9].x;
524 to[stride*10].x = aux[10].x;
525 to[stride*11].x = aux[11].x;
526 to[stride*12].x = aux[12].x;
527 to[stride*13].x = aux[13].x;
528 to[stride*14].x = aux[14].x;
529 to[stride*15].x = aux[15].x;
530}
531
532EIGEN_STRONG_INLINE void
533ptranspose(PacketBlock<Packet16h,16>& kernel) {
534 __m256i a = kernel.packet[0].x;
535 __m256i b = kernel.packet[1].x;
536 __m256i c = kernel.packet[2].x;
537 __m256i d = kernel.packet[3].x;
538 __m256i e = kernel.packet[4].x;
539 __m256i f = kernel.packet[5].x;
540 __m256i g = kernel.packet[6].x;
541 __m256i h = kernel.packet[7].x;
542 __m256i i = kernel.packet[8].x;
543 __m256i j = kernel.packet[9].x;
544 __m256i k = kernel.packet[10].x;
545 __m256i l = kernel.packet[11].x;
546 __m256i m = kernel.packet[12].x;
547 __m256i n = kernel.packet[13].x;
548 __m256i o = kernel.packet[14].x;
549 __m256i p = kernel.packet[15].x;
550
551 __m256i ab_07 = _mm256_unpacklo_epi16(a, b);
552 __m256i cd_07 = _mm256_unpacklo_epi16(c, d);
553 __m256i ef_07 = _mm256_unpacklo_epi16(e, f);
554 __m256i gh_07 = _mm256_unpacklo_epi16(g, h);
555 __m256i ij_07 = _mm256_unpacklo_epi16(i, j);
556 __m256i kl_07 = _mm256_unpacklo_epi16(k, l);
557 __m256i mn_07 = _mm256_unpacklo_epi16(m, n);
558 __m256i op_07 = _mm256_unpacklo_epi16(o, p);
559
560 __m256i ab_8f = _mm256_unpackhi_epi16(a, b);
561 __m256i cd_8f = _mm256_unpackhi_epi16(c, d);
562 __m256i ef_8f = _mm256_unpackhi_epi16(e, f);
563 __m256i gh_8f = _mm256_unpackhi_epi16(g, h);
564 __m256i ij_8f = _mm256_unpackhi_epi16(i, j);
565 __m256i kl_8f = _mm256_unpackhi_epi16(k, l);
566 __m256i mn_8f = _mm256_unpackhi_epi16(m, n);
567 __m256i op_8f = _mm256_unpackhi_epi16(o, p);
568
569 __m256i abcd_03 = _mm256_unpacklo_epi32(ab_07, cd_07);
570 __m256i abcd_47 = _mm256_unpackhi_epi32(ab_07, cd_07);
571 __m256i efgh_03 = _mm256_unpacklo_epi32(ef_07, gh_07);
572 __m256i efgh_47 = _mm256_unpackhi_epi32(ef_07, gh_07);
573 __m256i ijkl_03 = _mm256_unpacklo_epi32(ij_07, kl_07);
574 __m256i ijkl_47 = _mm256_unpackhi_epi32(ij_07, kl_07);
575 __m256i mnop_03 = _mm256_unpacklo_epi32(mn_07, op_07);
576 __m256i mnop_47 = _mm256_unpackhi_epi32(mn_07, op_07);
577
578 __m256i abcd_8b = _mm256_unpacklo_epi32(ab_8f, cd_8f);
579 __m256i abcd_cf = _mm256_unpackhi_epi32(ab_8f, cd_8f);
580 __m256i efgh_8b = _mm256_unpacklo_epi32(ef_8f, gh_8f);
581 __m256i efgh_cf = _mm256_unpackhi_epi32(ef_8f, gh_8f);
582 __m256i ijkl_8b = _mm256_unpacklo_epi32(ij_8f, kl_8f);
583 __m256i ijkl_cf = _mm256_unpackhi_epi32(ij_8f, kl_8f);
584 __m256i mnop_8b = _mm256_unpacklo_epi32(mn_8f, op_8f);
585 __m256i mnop_cf = _mm256_unpackhi_epi32(mn_8f, op_8f);
586
587 __m256i abcdefgh_01 = _mm256_unpacklo_epi64(abcd_03, efgh_03);
588 __m256i abcdefgh_23 = _mm256_unpackhi_epi64(abcd_03, efgh_03);
589 __m256i ijklmnop_01 = _mm256_unpacklo_epi64(ijkl_03, mnop_03);
590 __m256i ijklmnop_23 = _mm256_unpackhi_epi64(ijkl_03, mnop_03);
591 __m256i abcdefgh_45 = _mm256_unpacklo_epi64(abcd_47, efgh_47);
592 __m256i abcdefgh_67 = _mm256_unpackhi_epi64(abcd_47, efgh_47);
593 __m256i ijklmnop_45 = _mm256_unpacklo_epi64(ijkl_47, mnop_47);
594 __m256i ijklmnop_67 = _mm256_unpackhi_epi64(ijkl_47, mnop_47);
595 __m256i abcdefgh_89 = _mm256_unpacklo_epi64(abcd_8b, efgh_8b);
596 __m256i abcdefgh_ab = _mm256_unpackhi_epi64(abcd_8b, efgh_8b);
597 __m256i ijklmnop_89 = _mm256_unpacklo_epi64(ijkl_8b, mnop_8b);
598 __m256i ijklmnop_ab = _mm256_unpackhi_epi64(ijkl_8b, mnop_8b);
599 __m256i abcdefgh_cd = _mm256_unpacklo_epi64(abcd_cf, efgh_cf);
600 __m256i abcdefgh_ef = _mm256_unpackhi_epi64(abcd_cf, efgh_cf);
601 __m256i ijklmnop_cd = _mm256_unpacklo_epi64(ijkl_cf, mnop_cf);
602 __m256i ijklmnop_ef = _mm256_unpackhi_epi64(ijkl_cf, mnop_cf);
603
604 // NOTE: no unpacklo/hi instr in this case, so using permute instr.
605 __m256i a_p_0 = _mm256_permute2x128_si256(abcdefgh_01, ijklmnop_01, 0x20);
606 __m256i a_p_1 = _mm256_permute2x128_si256(abcdefgh_01, ijklmnop_01, 0x31);
607 __m256i a_p_2 = _mm256_permute2x128_si256(abcdefgh_23, ijklmnop_23, 0x20);
608 __m256i a_p_3 = _mm256_permute2x128_si256(abcdefgh_23, ijklmnop_23, 0x31);
609 __m256i a_p_4 = _mm256_permute2x128_si256(abcdefgh_45, ijklmnop_45, 0x20);
610 __m256i a_p_5 = _mm256_permute2x128_si256(abcdefgh_45, ijklmnop_45, 0x31);
611 __m256i a_p_6 = _mm256_permute2x128_si256(abcdefgh_67, ijklmnop_67, 0x20);
612 __m256i a_p_7 = _mm256_permute2x128_si256(abcdefgh_67, ijklmnop_67, 0x31);
613 __m256i a_p_8 = _mm256_permute2x128_si256(abcdefgh_89, ijklmnop_89, 0x20);
614 __m256i a_p_9 = _mm256_permute2x128_si256(abcdefgh_89, ijklmnop_89, 0x31);
615 __m256i a_p_a = _mm256_permute2x128_si256(abcdefgh_ab, ijklmnop_ab, 0x20);
616 __m256i a_p_b = _mm256_permute2x128_si256(abcdefgh_ab, ijklmnop_ab, 0x31);
617 __m256i a_p_c = _mm256_permute2x128_si256(abcdefgh_cd, ijklmnop_cd, 0x20);
618 __m256i a_p_d = _mm256_permute2x128_si256(abcdefgh_cd, ijklmnop_cd, 0x31);
619 __m256i a_p_e = _mm256_permute2x128_si256(abcdefgh_ef, ijklmnop_ef, 0x20);
620 __m256i a_p_f = _mm256_permute2x128_si256(abcdefgh_ef, ijklmnop_ef, 0x31);
621
622 kernel.packet[0].x = a_p_0;
623 kernel.packet[1].x = a_p_1;
624 kernel.packet[2].x = a_p_2;
625 kernel.packet[3].x = a_p_3;
626 kernel.packet[4].x = a_p_4;
627 kernel.packet[5].x = a_p_5;
628 kernel.packet[6].x = a_p_6;
629 kernel.packet[7].x = a_p_7;
630 kernel.packet[8].x = a_p_8;
631 kernel.packet[9].x = a_p_9;
632 kernel.packet[10].x = a_p_a;
633 kernel.packet[11].x = a_p_b;
634 kernel.packet[12].x = a_p_c;
635 kernel.packet[13].x = a_p_d;
636 kernel.packet[14].x = a_p_e;
637 kernel.packet[15].x = a_p_f;
638}
639
640EIGEN_STRONG_INLINE void
641ptranspose(PacketBlock<Packet16h,8>& kernel) {
642 EIGEN_ALIGN64 half in[8][16];
643 pstore<half>(in[0], kernel.packet[0]);
644 pstore<half>(in[1], kernel.packet[1]);
645 pstore<half>(in[2], kernel.packet[2]);
646 pstore<half>(in[3], kernel.packet[3]);
647 pstore<half>(in[4], kernel.packet[4]);
648 pstore<half>(in[5], kernel.packet[5]);
649 pstore<half>(in[6], kernel.packet[6]);
650 pstore<half>(in[7], kernel.packet[7]);
651
652 EIGEN_ALIGN64 half out[8][16];
653
654 for (int i = 0; i < 8; ++i) {
655 for (int j = 0; j < 8; ++j) {
656 out[i][j] = in[j][2*i];
657 }
658 for (int j = 0; j < 8; ++j) {
659 out[i][j+8] = in[j][2*i+1];
660 }
661 }
662
663 kernel.packet[0] = pload<Packet16h>(out[0]);
664 kernel.packet[1] = pload<Packet16h>(out[1]);
665 kernel.packet[2] = pload<Packet16h>(out[2]);
666 kernel.packet[3] = pload<Packet16h>(out[3]);
667 kernel.packet[4] = pload<Packet16h>(out[4]);
668 kernel.packet[5] = pload<Packet16h>(out[5]);
669 kernel.packet[6] = pload<Packet16h>(out[6]);
670 kernel.packet[7] = pload<Packet16h>(out[7]);
671}
672
673EIGEN_STRONG_INLINE void
674ptranspose(PacketBlock<Packet16h,4>& kernel) {
675 EIGEN_ALIGN64 half in[4][16];
676 pstore<half>(in[0], kernel.packet[0]);
677 pstore<half>(in[1], kernel.packet[1]);
678 pstore<half>(in[2], kernel.packet[2]);
679 pstore<half>(in[3], kernel.packet[3]);
680
681 EIGEN_ALIGN64 half out[4][16];
682
683 for (int i = 0; i < 4; ++i) {
684 for (int j = 0; j < 4; ++j) {
685 out[i][j] = in[j][4*i];
686 }
687 for (int j = 0; j < 4; ++j) {
688 out[i][j+4] = in[j][4*i+1];
689 }
690 for (int j = 0; j < 4; ++j) {
691 out[i][j+8] = in[j][4*i+2];
692 }
693 for (int j = 0; j < 4; ++j) {
694 out[i][j+12] = in[j][4*i+3];
695 }
696 }
697
698 kernel.packet[0] = pload<Packet16h>(out[0]);
699 kernel.packet[1] = pload<Packet16h>(out[1]);
700 kernel.packet[2] = pload<Packet16h>(out[2]);
701 kernel.packet[3] = pload<Packet16h>(out[3]);
702}
703
704
705#elif defined EIGEN_VECTORIZE_AVX
706
707typedef struct {
708 __m128i x;
709} Packet8h;
710
711
712template<> struct is_arithmetic<Packet8h> { enum { value = true }; };
713
714template <>
715struct packet_traits<Eigen::half> : default_packet_traits {
716 typedef Packet8h type;
717 // There is no half-size packet for Packet8h.
718 typedef Packet8h half;
719 enum {
720 Vectorizable = 1,
721 AlignedOnScalar = 1,
722 size = 8,
723 HasHalfPacket = 0,
724 HasAdd = 0,
725 HasSub = 0,
726 HasMul = 0,
727 HasNegate = 0,
728 HasAbs = 0,
729 HasAbs2 = 0,
730 HasMin = 0,
731 HasMax = 0,
732 HasConj = 0,
733 HasSetLinear = 0,
734 HasDiv = 0,
735 HasSqrt = 0,
736 HasRsqrt = 0,
737 HasExp = 0,
738 HasLog = 0,
739 HasBlend = 0
740 };
741};
742
743
744template<> struct unpacket_traits<Packet8h> { typedef Eigen::half type; enum {size=8, alignment=Aligned16}; typedef Packet8h half; };
745
746template<> EIGEN_STRONG_INLINE Packet8h pset1<Packet8h>(const Eigen::half& from) {
747 Packet8h result;
748 result.x = _mm_set1_epi16(from.x);
749 return result;
750}
751
752template<> EIGEN_STRONG_INLINE Eigen::half pfirst<Packet8h>(const Packet8h& from) {
753 return half_impl::raw_uint16_to_half(static_cast<unsigned short>(_mm_extract_epi16(from.x, 0)));
754}
755
756template<> EIGEN_STRONG_INLINE Packet8h pload<Packet8h>(const Eigen::half* from) {
757 Packet8h result;
758 result.x = _mm_load_si128(reinterpret_cast<const __m128i*>(from));
759 return result;
760}
761
762template<> EIGEN_STRONG_INLINE Packet8h ploadu<Packet8h>(const Eigen::half* from) {
763 Packet8h result;
764 result.x = _mm_loadu_si128(reinterpret_cast<const __m128i*>(from));
765 return result;
766}
767
768template<> EIGEN_STRONG_INLINE void pstore<Eigen::half>(Eigen::half* to, const Packet8h& from) {
769 _mm_store_si128(reinterpret_cast<__m128i*>(to), from.x);
770}
771
772template<> EIGEN_STRONG_INLINE void pstoreu<Eigen::half>(Eigen::half* to, const Packet8h& from) {
773 _mm_storeu_si128(reinterpret_cast<__m128i*>(to), from.x);
774}
775
776template<> EIGEN_STRONG_INLINE Packet8h
777ploadquad<Packet8h>(const Eigen::half* from) {
778 Packet8h result;
779 unsigned short a = from[0].x;
780 unsigned short b = from[1].x;
781 result.x = _mm_set_epi16(b, b, b, b, a, a, a, a);
782 return result;
783}
784
785EIGEN_STRONG_INLINE Packet8f half2float(const Packet8h& a) {
786#ifdef EIGEN_HAS_FP16_C
787 return _mm256_cvtph_ps(a.x);
788#else
789 EIGEN_ALIGN32 Eigen::half aux[8];
790 pstore(aux, a);
791 float f0(aux[0]);
792 float f1(aux[1]);
793 float f2(aux[2]);
794 float f3(aux[3]);
795 float f4(aux[4]);
796 float f5(aux[5]);
797 float f6(aux[6]);
798 float f7(aux[7]);
799
800 return _mm256_set_ps(f7, f6, f5, f4, f3, f2, f1, f0);
801#endif
802}
803
804EIGEN_STRONG_INLINE Packet8h float2half(const Packet8f& a) {
805#ifdef EIGEN_HAS_FP16_C
806 Packet8h result;
807 result.x = _mm256_cvtps_ph(a, _MM_FROUND_TO_NEAREST_INT|_MM_FROUND_NO_EXC);
808 return result;
809#else
810 EIGEN_ALIGN32 float aux[8];
811 pstore(aux, a);
812 Eigen::half h0(aux[0]);
813 Eigen::half h1(aux[1]);
814 Eigen::half h2(aux[2]);
815 Eigen::half h3(aux[3]);
816 Eigen::half h4(aux[4]);
817 Eigen::half h5(aux[5]);
818 Eigen::half h6(aux[6]);
819 Eigen::half h7(aux[7]);
820
821 Packet8h result;
822 result.x = _mm_set_epi16(h7.x, h6.x, h5.x, h4.x, h3.x, h2.x, h1.x, h0.x);
823 return result;
824#endif
825}
826
827template<> EIGEN_STRONG_INLINE Packet8h pconj(const Packet8h& a) { return a; }
828
829template<> EIGEN_STRONG_INLINE Packet8h padd<Packet8h>(const Packet8h& a, const Packet8h& b) {
830 Packet8f af = half2float(a);
831 Packet8f bf = half2float(b);
832 Packet8f rf = padd(af, bf);
833 return float2half(rf);
834}
835
836template<> EIGEN_STRONG_INLINE Packet8h pmul<Packet8h>(const Packet8h& a, const Packet8h& b) {
837 Packet8f af = half2float(a);
838 Packet8f bf = half2float(b);
839 Packet8f rf = pmul(af, bf);
840 return float2half(rf);
841}
842
843template<> EIGEN_STRONG_INLINE Packet8h pgather<Eigen::half, Packet8h>(const Eigen::half* from, Index stride)
844{
845 Packet8h result;
846 result.x = _mm_set_epi16(from[7*stride].x, from[6*stride].x, from[5*stride].x, from[4*stride].x, from[3*stride].x, from[2*stride].x, from[1*stride].x, from[0*stride].x);
847 return result;
848}
849
850template<> EIGEN_STRONG_INLINE void pscatter<Eigen::half, Packet8h>(Eigen::half* to, const Packet8h& from, Index stride)
851{
852 EIGEN_ALIGN32 Eigen::half aux[8];
853 pstore(aux, from);
854 to[stride*0].x = aux[0].x;
855 to[stride*1].x = aux[1].x;
856 to[stride*2].x = aux[2].x;
857 to[stride*3].x = aux[3].x;
858 to[stride*4].x = aux[4].x;
859 to[stride*5].x = aux[5].x;
860 to[stride*6].x = aux[6].x;
861 to[stride*7].x = aux[7].x;
862}
863
864template<> EIGEN_STRONG_INLINE Eigen::half predux<Packet8h>(const Packet8h& a) {
865 Packet8f af = half2float(a);
866 float reduced = predux<Packet8f>(af);
867 return Eigen::half(reduced);
868}
869
870template<> EIGEN_STRONG_INLINE Eigen::half predux_max<Packet8h>(const Packet8h& a) {
871 Packet8f af = half2float(a);
872 float reduced = predux_max<Packet8f>(af);
873 return Eigen::half(reduced);
874}
875
876template<> EIGEN_STRONG_INLINE Eigen::half predux_min<Packet8h>(const Packet8h& a) {
877 Packet8f af = half2float(a);
878 float reduced = predux_min<Packet8f>(af);
879 return Eigen::half(reduced);
880}
881
882template<> EIGEN_STRONG_INLINE Eigen::half predux_mul<Packet8h>(const Packet8h& a) {
883 Packet8f af = half2float(a);
884 float reduced = predux_mul<Packet8f>(af);
885 return Eigen::half(reduced);
886}
887
888EIGEN_STRONG_INLINE void
889ptranspose(PacketBlock<Packet8h,8>& kernel) {
890 __m128i a = kernel.packet[0].x;
891 __m128i b = kernel.packet[1].x;
892 __m128i c = kernel.packet[2].x;
893 __m128i d = kernel.packet[3].x;
894 __m128i e = kernel.packet[4].x;
895 __m128i f = kernel.packet[5].x;
896 __m128i g = kernel.packet[6].x;
897 __m128i h = kernel.packet[7].x;
898
899 __m128i a03b03 = _mm_unpacklo_epi16(a, b);
900 __m128i c03d03 = _mm_unpacklo_epi16(c, d);
901 __m128i e03f03 = _mm_unpacklo_epi16(e, f);
902 __m128i g03h03 = _mm_unpacklo_epi16(g, h);
903 __m128i a47b47 = _mm_unpackhi_epi16(a, b);
904 __m128i c47d47 = _mm_unpackhi_epi16(c, d);
905 __m128i e47f47 = _mm_unpackhi_epi16(e, f);
906 __m128i g47h47 = _mm_unpackhi_epi16(g, h);
907
908 __m128i a01b01c01d01 = _mm_unpacklo_epi32(a03b03, c03d03);
909 __m128i a23b23c23d23 = _mm_unpackhi_epi32(a03b03, c03d03);
910 __m128i e01f01g01h01 = _mm_unpacklo_epi32(e03f03, g03h03);
911 __m128i e23f23g23h23 = _mm_unpackhi_epi32(e03f03, g03h03);
912 __m128i a45b45c45d45 = _mm_unpacklo_epi32(a47b47, c47d47);
913 __m128i a67b67c67d67 = _mm_unpackhi_epi32(a47b47, c47d47);
914 __m128i e45f45g45h45 = _mm_unpacklo_epi32(e47f47, g47h47);
915 __m128i e67f67g67h67 = _mm_unpackhi_epi32(e47f47, g47h47);
916
917 __m128i a0b0c0d0e0f0g0h0 = _mm_unpacklo_epi64(a01b01c01d01, e01f01g01h01);
918 __m128i a1b1c1d1e1f1g1h1 = _mm_unpackhi_epi64(a01b01c01d01, e01f01g01h01);
919 __m128i a2b2c2d2e2f2g2h2 = _mm_unpacklo_epi64(a23b23c23d23, e23f23g23h23);
920 __m128i a3b3c3d3e3f3g3h3 = _mm_unpackhi_epi64(a23b23c23d23, e23f23g23h23);
921 __m128i a4b4c4d4e4f4g4h4 = _mm_unpacklo_epi64(a45b45c45d45, e45f45g45h45);
922 __m128i a5b5c5d5e5f5g5h5 = _mm_unpackhi_epi64(a45b45c45d45, e45f45g45h45);
923 __m128i a6b6c6d6e6f6g6h6 = _mm_unpacklo_epi64(a67b67c67d67, e67f67g67h67);
924 __m128i a7b7c7d7e7f7g7h7 = _mm_unpackhi_epi64(a67b67c67d67, e67f67g67h67);
925
926 kernel.packet[0].x = a0b0c0d0e0f0g0h0;
927 kernel.packet[1].x = a1b1c1d1e1f1g1h1;
928 kernel.packet[2].x = a2b2c2d2e2f2g2h2;
929 kernel.packet[3].x = a3b3c3d3e3f3g3h3;
930 kernel.packet[4].x = a4b4c4d4e4f4g4h4;
931 kernel.packet[5].x = a5b5c5d5e5f5g5h5;
932 kernel.packet[6].x = a6b6c6d6e6f6g6h6;
933 kernel.packet[7].x = a7b7c7d7e7f7g7h7;
934}
935
936EIGEN_STRONG_INLINE void
937ptranspose(PacketBlock<Packet8h,4>& kernel) {
938 EIGEN_ALIGN32 Eigen::half in[4][8];
939 pstore<Eigen::half>(in[0], kernel.packet[0]);
940 pstore<Eigen::half>(in[1], kernel.packet[1]);
941 pstore<Eigen::half>(in[2], kernel.packet[2]);
942 pstore<Eigen::half>(in[3], kernel.packet[3]);
943
944 EIGEN_ALIGN32 Eigen::half out[4][8];
945
946 for (int i = 0; i < 4; ++i) {
947 for (int j = 0; j < 4; ++j) {
948 out[i][j] = in[j][2*i];
949 }
950 for (int j = 0; j < 4; ++j) {
951 out[i][j+4] = in[j][2*i+1];
952 }
953 }
954
955 kernel.packet[0] = pload<Packet8h>(out[0]);
956 kernel.packet[1] = pload<Packet8h>(out[1]);
957 kernel.packet[2] = pload<Packet8h>(out[2]);
958 kernel.packet[3] = pload<Packet8h>(out[3]);
959}
960
961
962// Disable the following code since it's broken on too many platforms / compilers.
963//#elif defined(EIGEN_VECTORIZE_SSE) && (!EIGEN_ARCH_x86_64) && (!EIGEN_COMP_MSVC)
964#elif 0
965
966typedef struct {
967 __m64 x;
968} Packet4h;
969
970
971template<> struct is_arithmetic<Packet4h> { enum { value = true }; };
972
973template <>
974struct packet_traits<Eigen::half> : default_packet_traits {
975 typedef Packet4h type;
976 // There is no half-size packet for Packet4h.
977 typedef Packet4h half;
978 enum {
979 Vectorizable = 1,
980 AlignedOnScalar = 1,
981 size = 4,
982 HasHalfPacket = 0,
983 HasAdd = 0,
984 HasSub = 0,
985 HasMul = 0,
986 HasNegate = 0,
987 HasAbs = 0,
988 HasAbs2 = 0,
989 HasMin = 0,
990 HasMax = 0,
991 HasConj = 0,
992 HasSetLinear = 0,
993 HasDiv = 0,
994 HasSqrt = 0,
995 HasRsqrt = 0,
996 HasExp = 0,
997 HasLog = 0,
998 HasBlend = 0
999 };
1000};
1001
1002
1003template<> struct unpacket_traits<Packet4h> { typedef Eigen::half type; enum {size=4, alignment=Aligned16}; typedef Packet4h half; };
1004
1005template<> EIGEN_STRONG_INLINE Packet4h pset1<Packet4h>(const Eigen::half& from) {
1006 Packet4h result;
1007 result.x = _mm_set1_pi16(from.x);
1008 return result;
1009}
1010
1011template<> EIGEN_STRONG_INLINE Eigen::half pfirst<Packet4h>(const Packet4h& from) {
1012 return half_impl::raw_uint16_to_half(static_cast<unsigned short>(_mm_cvtsi64_si32(from.x)));
1013}
1014
1015template<> EIGEN_STRONG_INLINE Packet4h pconj(const Packet4h& a) { return a; }
1016
1017template<> EIGEN_STRONG_INLINE Packet4h padd<Packet4h>(const Packet4h& a, const Packet4h& b) {
1018 __int64_t a64 = _mm_cvtm64_si64(a.x);
1019 __int64_t b64 = _mm_cvtm64_si64(b.x);
1020
1021 Eigen::half h[4];
1022
1023 Eigen::half ha = half_impl::raw_uint16_to_half(static_cast<unsigned short>(a64));
1024 Eigen::half hb = half_impl::raw_uint16_to_half(static_cast<unsigned short>(b64));
1025 h[0] = ha + hb;
1026 ha = half_impl::raw_uint16_to_half(static_cast<unsigned short>(a64 >> 16));
1027 hb = half_impl::raw_uint16_to_half(static_cast<unsigned short>(b64 >> 16));
1028 h[1] = ha + hb;
1029 ha = half_impl::raw_uint16_to_half(static_cast<unsigned short>(a64 >> 32));
1030 hb = half_impl::raw_uint16_to_half(static_cast<unsigned short>(b64 >> 32));
1031 h[2] = ha + hb;
1032 ha = half_impl::raw_uint16_to_half(static_cast<unsigned short>(a64 >> 48));
1033 hb = half_impl::raw_uint16_to_half(static_cast<unsigned short>(b64 >> 48));
1034 h[3] = ha + hb;
1035 Packet4h result;
1036 result.x = _mm_set_pi16(h[3].x, h[2].x, h[1].x, h[0].x);
1037 return result;
1038}
1039
1040template<> EIGEN_STRONG_INLINE Packet4h pmul<Packet4h>(const Packet4h& a, const Packet4h& b) {
1041 __int64_t a64 = _mm_cvtm64_si64(a.x);
1042 __int64_t b64 = _mm_cvtm64_si64(b.x);
1043
1044 Eigen::half h[4];
1045
1046 Eigen::half ha = half_impl::raw_uint16_to_half(static_cast<unsigned short>(a64));
1047 Eigen::half hb = half_impl::raw_uint16_to_half(static_cast<unsigned short>(b64));
1048 h[0] = ha * hb;
1049 ha = half_impl::raw_uint16_to_half(static_cast<unsigned short>(a64 >> 16));
1050 hb = half_impl::raw_uint16_to_half(static_cast<unsigned short>(b64 >> 16));
1051 h[1] = ha * hb;
1052 ha = half_impl::raw_uint16_to_half(static_cast<unsigned short>(a64 >> 32));
1053 hb = half_impl::raw_uint16_to_half(static_cast<unsigned short>(b64 >> 32));
1054 h[2] = ha * hb;
1055 ha = half_impl::raw_uint16_to_half(static_cast<unsigned short>(a64 >> 48));
1056 hb = half_impl::raw_uint16_to_half(static_cast<unsigned short>(b64 >> 48));
1057 h[3] = ha * hb;
1058 Packet4h result;
1059 result.x = _mm_set_pi16(h[3].x, h[2].x, h[1].x, h[0].x);
1060 return result;
1061}
1062
1063template<> EIGEN_STRONG_INLINE Packet4h pload<Packet4h>(const Eigen::half* from) {
1064 Packet4h result;
1065 result.x = _mm_cvtsi64_m64(*reinterpret_cast<const __int64_t*>(from));
1066 return result;
1067}
1068
1069template<> EIGEN_STRONG_INLINE Packet4h ploadu<Packet4h>(const Eigen::half* from) {
1070 Packet4h result;
1071 result.x = _mm_cvtsi64_m64(*reinterpret_cast<const __int64_t*>(from));
1072 return result;
1073}
1074
1075template<> EIGEN_STRONG_INLINE void pstore<Eigen::half>(Eigen::half* to, const Packet4h& from) {
1076 __int64_t r = _mm_cvtm64_si64(from.x);
1077 *(reinterpret_cast<__int64_t*>(to)) = r;
1078}
1079
1080template<> EIGEN_STRONG_INLINE void pstoreu<Eigen::half>(Eigen::half* to, const Packet4h& from) {
1081 __int64_t r = _mm_cvtm64_si64(from.x);
1082 *(reinterpret_cast<__int64_t*>(to)) = r;
1083}
1084
1085template<> EIGEN_STRONG_INLINE Packet4h
1086ploadquad<Packet4h>(const Eigen::half* from) {
1087 return pset1<Packet4h>(*from);
1088}
1089
1090template<> EIGEN_STRONG_INLINE Packet4h pgather<Eigen::half, Packet4h>(const Eigen::half* from, Index stride)
1091{
1092 Packet4h result;
1093 result.x = _mm_set_pi16(from[3*stride].x, from[2*stride].x, from[1*stride].x, from[0*stride].x);
1094 return result;
1095}
1096
1097template<> EIGEN_STRONG_INLINE void pscatter<Eigen::half, Packet4h>(Eigen::half* to, const Packet4h& from, Index stride)
1098{
1099 __int64_t a = _mm_cvtm64_si64(from.x);
1100 to[stride*0].x = static_cast<unsigned short>(a);
1101 to[stride*1].x = static_cast<unsigned short>(a >> 16);
1102 to[stride*2].x = static_cast<unsigned short>(a >> 32);
1103 to[stride*3].x = static_cast<unsigned short>(a >> 48);
1104}
1105
1106EIGEN_STRONG_INLINE void
1107ptranspose(PacketBlock<Packet4h,4>& kernel) {
1108 __m64 T0 = _mm_unpacklo_pi16(kernel.packet[0].x, kernel.packet[1].x);
1109 __m64 T1 = _mm_unpacklo_pi16(kernel.packet[2].x, kernel.packet[3].x);
1110 __m64 T2 = _mm_unpackhi_pi16(kernel.packet[0].x, kernel.packet[1].x);
1111 __m64 T3 = _mm_unpackhi_pi16(kernel.packet[2].x, kernel.packet[3].x);
1112
1113 kernel.packet[0].x = _mm_unpacklo_pi32(T0, T1);
1114 kernel.packet[1].x = _mm_unpackhi_pi32(T0, T1);
1115 kernel.packet[2].x = _mm_unpacklo_pi32(T2, T3);
1116 kernel.packet[3].x = _mm_unpackhi_pi32(T2, T3);
1117}
1118
1119#endif
1120
1121}
1122}
1123
1124#endif // EIGEN_PACKET_MATH_HALF_CUDA_H
1125