Skip to content

Commit fc78c90

Browse files
BaiXilinBaiXilin
committed
header file comment format
1 parent cc40d7b commit fc78c90

File tree

1 file changed

+81
-87
lines changed

1 file changed

+81
-87
lines changed

‎clang/lib/Headers/avxvnniint16intrin.h‎

Lines changed: 81 additions & 87 deletions
Original file line numberDiff line numberDiff line change
@@ -15,11 +15,11 @@
1515
#ifndef __AVXVNNIINT16INTRIN_H
1616
#define __AVXVNNIINT16INTRIN_H
1717

18-
// clang-format off
1918
/// Multiply groups of 2 adjacent pairs of signed 16-bit integers in \a __A with
20-
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
21-
/// signed 16-bit results. Sum these 2 results with the corresponding
22-
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
19+
/// corresponding unsigned 16-bit integers in \a __B, producing 2
20+
/// intermediate signed 16-bit results. Sum these 2 results with the
21+
/// corresponding 32-bit integer in \a __W, and store the packed 32-bit
22+
/// results in \a dst.
2323
///
2424
/// \headerfile <immintrin.h>
2525
///
@@ -41,21 +41,21 @@
4141
/// \code{.operation}
4242
/// FOR j := 0 to 3
4343
/// tmp1.dword := SignExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
44-
/// tmp2.dword := SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
44+
/// tmp2.dword :=
45+
/// SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
4546
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
4647
/// ENDFOR
4748
/// dst[MAX:128] := 0
4849
/// \endcode
49-
// clang-format on
5050
#define _mm_dpwsud_epi32(__W, __A, __B) \
5151
((__m128i)__builtin_ia32_vpdpwsud128((__v4si)(__W), (__v8hi)(__A), \
5252
(__v8hu)(__B)))
5353

54-
// clang-format off
5554
/// Multiply groups of 2 adjacent pairs of signed 16-bit integers in \a __A with
56-
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
57-
/// signed 16-bit results. Sum these 2 results with the corresponding
58-
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
55+
/// corresponding unsigned 16-bit integers in \a __B, producing 2
56+
/// intermediate signed 16-bit results. Sum these 2 results with the
57+
/// corresponding 32-bit integer in \a __W, and store the packed 32-bit
58+
/// results in \a dst.
5959
///
6060
/// \headerfile <immintrin.h>
6161
///
@@ -77,22 +77,21 @@
7777
/// \code{.operation}
7878
/// FOR j := 0 to 7
7979
/// tmp1.dword := SignExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
80-
/// tmp2.dword := SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
81-
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
80+
/// tmp2.dword :=
81+
/// SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
82+
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
8283
/// ENDFOR
8384
/// dst[MAX:256] := 0
8485
/// \endcode
85-
// clang-format on
8686
#define _mm256_dpwsud_epi32(__W, __A, __B) \
8787
((__m256i)__builtin_ia32_vpdpwsud256((__v8si)(__W), (__v16hi)(__A), \
8888
(__v16hu)(__B)))
8989

90-
// clang-format off
9190
/// Multiply groups of 2 adjacent pairs of signed 16-bit integers in \a __A with
92-
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
93-
/// signed 16-bit results. Sum these 2 results with the corresponding
94-
/// 32-bit integer in \a __W with signed saturation, and store the packed
95-
/// 32-bit results in \a dst.
91+
/// corresponding unsigned 16-bit integers in \a __B, producing 2
92+
/// intermediate signed 16-bit results. Sum these 2 results with the
93+
/// corresponding 32-bit integer in \a __W with signed saturation, and store
94+
/// the packed 32-bit results in \a dst.
9695
///
9796
/// \headerfile <immintrin.h>
9897
///
@@ -114,23 +113,22 @@
114113
/// \code{.operation}
115114
/// FOR j := 0 to 3
116115
/// tmp1.dword := SignExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
117-
/// tmp2.dword := SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
116+
/// tmp2.dword :=
117+
/// SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
118118
/// dst.dword[j] := SIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
119119
/// ENDFOR
120120
/// dst[MAX:128] := 0
121121
/// \endcode
122-
// clang-format on
123122
/// Multiply groups of 2 adjacent pairs of signed 16-bit integers in \a __A with
124123
#define _mm_dpwsuds_epi32(__W, __A, __B) \
125124
((__m128i)__builtin_ia32_vpdpwsuds128((__v4si)(__W), (__v8hi)(__A), \
126125
(__v8hu)(__B)))
127126

128-
// clang-format off
129127
/// Multiply groups of 2 adjacent pairs of signed 16-bit integers in \a __A with
130-
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
131-
/// signed 16-bit results. Sum these 2 results with the corresponding
132-
/// 32-bit integer in \a __W with signed saturation, and store the packed
133-
/// 32-bit results in \a dst.
128+
/// corresponding unsigned 16-bit integers in \a __B, producing 2
129+
/// intermediate signed 16-bit results. Sum these 2 results with the
130+
/// corresponding 32-bit integer in \a __W with signed saturation, and store
131+
/// the packed 32-bit results in \a dst.
134132
///
135133
/// \headerfile <immintrin.h>
136134
///
@@ -152,21 +150,21 @@
152150
/// \code{.operation}
153151
/// FOR j := 0 to 7
154152
/// tmp1.dword := SignExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
155-
/// tmp2.dword := SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
153+
/// tmp2.dword :=
154+
/// SignExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
156155
/// dst.dword[j] := SIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
157156
/// ENDFOR
158157
/// dst[MAX:256] := 0
159158
/// \endcode
160-
// clang-format on
161159
#define _mm256_dpwsuds_epi32(__W, __A, __B) \
162160
((__m256i)__builtin_ia32_vpdpwsuds256((__v8si)(__W), (__v16hi)(__A), \
163161
(__v16hu)(__B)))
164162

165-
// clang-format off
166-
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
167-
/// corresponding signed 16-bit integers in \a __B, producing 2 intermediate
168-
/// signed 16-bit results. Sum these 2 results with the corresponding
169-
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
163+
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A
164+
/// with corresponding signed 16-bit integers in \a __B, producing 2
165+
/// intermediate signed 16-bit results. Sum these 2 results with the
166+
/// corresponding 32-bit integer in \a __W, and store the packed 32-bit
167+
/// results in \a dst.
170168
///
171169
/// \headerfile <immintrin.h>
172170
///
@@ -188,21 +186,21 @@
188186
/// \code{.operation}
189187
/// FOR j := 0 to 3
190188
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * SignExtend32(__B.word[2*j])
191-
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
189+
/// tmp2.dword :=
190+
/// ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
192191
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
193192
/// ENDFOR
194193
/// dst[MAX:128] := 0
195194
/// \endcode
196-
// clang-format on
197195
#define _mm_dpwusd_epi32(__W, __A, __B) \
198196
((__m128i)__builtin_ia32_vpdpwusd128((__v4si)(__W), (__v8hu)(__A), \
199197
(__v8hi)(__B)))
200198

201-
// clang-format off
202-
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
203-
/// corresponding signed 16-bit integers in \a __B, producing 2 intermediate
204-
/// signed 16-bit results. Sum these 2 results with the corresponding
205-
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
199+
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A
200+
/// with corresponding signed 16-bit integers in \a __B, producing 2
201+
/// intermediate signed 16-bit results. Sum these 2 results with the
202+
/// corresponding 32-bit integer in \a __W, and store the packed 32-bit
203+
/// results in \a dst.
206204
///
207205
/// \headerfile <immintrin.h>
208206
///
@@ -224,22 +222,21 @@
224222
/// \code{.operation}
225223
/// FOR j := 0 to 7
226224
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * SignExtend32(__B.word[2*j])
227-
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
225+
/// tmp2.dword :=
226+
/// ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
228227
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
229228
/// ENDFOR
230229
/// dst[MAX:256] := 0
231230
/// \endcode
232-
// clang-format on
233231
#define _mm256_dpwusd_epi32(__W, __A, __B) \
234232
((__m256i)__builtin_ia32_vpdpwusd256((__v8si)(__W), (__v16hu)(__A), \
235233
(__v16hi)(__B)))
236234

237-
// clang-format off
238-
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
239-
/// corresponding signed 16-bit integers in \a __B, producing 2 intermediate
240-
/// signed 16-bit results. Sum these 2 results with the corresponding
241-
/// 32-bit integer in \a __W with signed saturation, and store the packed
242-
/// 32-bit results in \a dst.
235+
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A
236+
/// with corresponding signed 16-bit integers in \a __B, producing 2
237+
/// intermediate signed 16-bit results. Sum these 2 results with the
238+
/// corresponding 32-bit integer in \a __W with signed saturation, and
239+
/// store the packed 32-bit results in \a dst.
243240
///
244241
/// \headerfile <immintrin.h>
245242
///
@@ -261,22 +258,21 @@
261258
/// \code{.operation}
262259
/// FOR j := 0 to 3
263260
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * SignExtend32(__B.word[2*j])
264-
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
261+
/// tmp2.dword :=
262+
/// ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
265263
/// dst.dword[j] := SIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
266264
/// ENDFOR
267265
/// dst[MAX:128] := 0
268266
/// \endcode
269-
// clang-format on
270267
#define _mm_dpwusds_epi32(__W, __A, __B) \
271268
((__m128i)__builtin_ia32_vpdpwusds128((__v4si)(__W), (__v8hu)(__A), \
272269
(__v8hi)(__B)))
273270

274-
// clang-format off
275-
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
276-
/// corresponding signed 16-bit integers in \a __B, producing 2 intermediate
277-
/// signed 16-bit results. Sum these 2 results with the corresponding
278-
/// 32-bit integer in \a __W with signed saturation, and store the packed
279-
/// 32-bit results in \a dst.
271+
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A
272+
/// with corresponding signed 16-bit integers in \a __B, producing 2
273+
/// intermediate signed 16-bit results. Sum these 2 results with the
274+
/// corresponding 32-bit integer in \a __W with signed saturation, and
275+
/// store the packed 32-bit results in \a dst.
280276
///
281277
/// \headerfile <immintrin.h>
282278
///
@@ -298,21 +294,21 @@
298294
/// \code{.operation}
299295
/// FOR j := 0 to 7
300296
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * SignExtend32(__B.word[2*j])
301-
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
297+
/// tmp2.dword :=
298+
/// ZeroExtend32(__A.word[2*j+1]) * SignExtend32(__B.word[2*j+1])
302299
/// dst.dword[j] := SIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
303300
/// ENDFOR
304301
/// dst[MAX:256] := 0
305302
/// \endcode
306-
// clang-format on
307303
#define _mm256_dpwusds_epi32(__W, __A, __B) \
308304
((__m256i)__builtin_ia32_vpdpwusds256((__v8si)(__W), (__v16hu)(__A), \
309305
(__v16hi)(__B)))
310306

311-
// clang-format off
312-
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
313-
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
314-
/// signed 16-bit results. Sum these 2 results with the corresponding
315-
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
307+
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A
308+
/// with corresponding unsigned 16-bit integers in \a __B, producing 2
309+
/// intermediate signed 16-bit results. Sum these 2 results with the
310+
/// corresponding 32-bit integer in \a __W, and store the packed 32-bit
311+
/// results in \a dst.
316312
///
317313
/// \headerfile <immintrin.h>
318314
///
@@ -334,21 +330,21 @@
334330
/// \code{.operation}
335331
/// FOR j := 0 to 3
336332
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
337-
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
333+
/// tmp2.dword :=
334+
/// ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
338335
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
339336
/// ENDFOR
340337
/// dst[MAX:128] := 0
341338
/// \endcode
342-
// clang-format on
343339
#define _mm_dpwuud_epi32(__W, __A, __B) \
344340
((__m128i)__builtin_ia32_vpdpwuud128((__v4si)(__W), (__v8hu)(__A), \
345341
(__v8hu)(__B)))
346342

347-
// clang-format off
348-
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
349-
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
350-
/// signed 16-bit results. Sum these 2 results with the corresponding
351-
/// 32-bit integer in \a __W, and store the packed 32-bit results in \a dst.
343+
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A
344+
/// with corresponding unsigned 16-bit integers in \a __B, producing 2
345+
/// intermediate signed 16-bit results. Sum these 2 results with the
346+
/// corresponding 32-bit integer in \a __W, and store the packed 32-bit
347+
/// results in \a dst.
352348
///
353349
/// \headerfile <immintrin.h>
354350
///
@@ -370,22 +366,21 @@
370366
/// \code{.operation}
371367
/// FOR j := 0 to 7
372368
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
373-
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
369+
/// tmp2.dword :=
370+
/// ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
374371
/// dst.dword[j] := __W.dword[j] + tmp1 + tmp2
375372
/// ENDFOR
376373
/// dst[MAX:256] := 0
377374
/// \endcode
378-
// clang-format on
379375
#define _mm256_dpwuud_epi32(__W, __A, __B) \
380376
((__m256i)__builtin_ia32_vpdpwuud256((__v8si)(__W), (__v16hu)(__A), \
381377
(__v16hu)(__B)))
382378

383-
// clang-format off
384-
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
385-
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
386-
/// signed 16-bit results. Sum these 2 results with the corresponding
387-
/// 32-bit integer in \a __W with signed saturation, and store the packed
388-
/// 32-bit results in \a dst.
379+
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A
380+
/// with corresponding unsigned 16-bit integers in \a __B, producing 2
381+
/// intermediate signed 16-bit results. Sum these 2 results with the
382+
/// corresponding 32-bit integer in \a __W with signed saturation, and store
383+
/// the packed 32-bit results in \a dst.
389384
///
390385
/// \headerfile <immintrin.h>
391386
///
@@ -407,22 +402,21 @@
407402
/// \code{.operation}
408403
/// FOR j := 0 to 3
409404
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
410-
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
405+
/// tmp2.dword :=
406+
/// ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
411407
/// dst.dword[j] := UNSIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
412408
/// ENDFOR
413409
/// dst[MAX:128] := 0
414410
/// \endcode
415-
// clang-format on
416411
#define _mm_dpwuuds_epi32(__W, __A, __B) \
417412
((__m128i)__builtin_ia32_vpdpwuuds128((__v4si)(__W), (__v8hu)(__A), \
418413
(__v8hu)(__B)))
419414

420-
// clang-format off
421-
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A with
422-
/// corresponding unsigned 16-bit integers in \a __B, producing 2 intermediate
423-
/// signed 16-bit results. Sum these 2 results with the corresponding
424-
/// 32-bit integer in \a __W with signed saturation, and store the packed
425-
/// 32-bit results in \a dst.
415+
/// Multiply groups of 2 adjacent pairs of unsigned 16-bit integers in \a __A
416+
/// with corresponding unsigned 16-bit integers in \a __B, producing 2
417+
/// intermediate signed 16-bit results. Sum these 2 results with the
418+
/// corresponding 32-bit integer in \a __W with signed saturation, and store
419+
/// the packed 32-bit results in \a dst.
426420
///
427421
/// \headerfile <immintrin.h>
428422
///
@@ -444,12 +438,12 @@
444438
/// \code{.operation}
445439
/// FOR j := 0 to 7
446440
/// tmp1.dword := ZeroExtend32(__A.word[2*j]) * ZeroExtend32(__B.word[2*j])
447-
/// tmp2.dword := ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
441+
/// tmp2.dword :=
442+
/// ZeroExtend32(__A.word[2*j+1]) * ZeroExtend32(__B.word[2*j+1])
448443
/// dst.dword[j] := UNSIGNED_DWORD_SATURATE(__W.dword[j] + tmp1 + tmp2)
449444
/// ENDFOR
450445
/// dst[MAX:256] := 0
451446
/// \endcode
452-
// clang-format on
453447
#define _mm256_dpwuuds_epi32(__W, __A, __B) \
454448
((__m256i)__builtin_ia32_vpdpwuuds256((__v8si)(__W), (__v16hu)(__A), \
455449
(__v16hu)(__B)))

0 commit comments

Comments
 (0)