1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
//! `x86` and `x86_64` intrinsics.

use crate::{intrinsics, marker::Sized, mem::transmute};

#[macro_use]
mod macros;

types! {
    /// 64-bit wide integer vector type, x86-specific
    ///
    /// This type is the same as the `__m64` type defined by Intel,
    /// representing a 64-bit SIMD register. Usage of this type typically
    /// corresponds to the `mmx` target feature.
    ///
    /// Internally this type may be viewed as:
    ///
    /// * `i8x8` - eight `i8` variables packed together
    /// * `i16x4` - four `i16` variables packed together
    /// * `i32x2` - two `i32` variables packed together
    ///
    /// (as well as unsigned versions). Each intrinsic may interpret the
    /// internal bits differently, check the documentation of the intrinsic
    /// to see how it's being used.
    ///
    /// Note that this means that an instance of `__m64` typically just means
    /// a "bag of bits" which is left up to interpretation at the point of use.
    ///
    /// Most intrinsics using `__m64` are prefixed with `_mm_` and the
    /// integer types tend to correspond to suffixes like "pi8" or "pi32" (not
    /// to be confused with "epiXX", used for `__m128i`).
    ///
    /// # Examples
    ///
    /// ```
    /// # #![feature(stdsimd, mmx_target_feature)]
    /// #[cfg(target_arch = "x86")]
    /// use std::arch::x86::*;
    /// #[cfg(target_arch = "x86_64")]
    /// use std::arch::x86_64::*;
    ///
    /// # fn main() {
    /// # #[target_feature(enable = "mmx")]
    /// # unsafe fn foo() {
    /// let all_bytes_zero = _mm_setzero_si64();
    /// let all_bytes_one = _mm_set1_pi8(1);
    /// let two_i32 = _mm_set_pi32(1, 2);
    /// # }
    /// # if is_x86_feature_detected!("mmx") { unsafe { foo() } }
    /// # }
    /// ```
    pub struct __m64(i64);

    /// 128-bit wide integer vector type, x86-specific
    ///
    /// This type is the same as the `__m128i` type defined by Intel,
    /// representing a 128-bit SIMD register. Usage of this type typically
    /// corresponds to the `sse` and up target features for x86/x86_64.
    ///
    /// Internally this type may be viewed as:
    ///
    /// * `i8x16` - sixteen `i8` variables packed together
    /// * `i16x8` - eight `i16` variables packed together
    /// * `i32x4` - four `i32` variables packed together
    /// * `i64x2` - two `i64` variables packed together
    ///
    /// (as well as unsigned versions). Each intrinsic may interpret the
    /// internal bits differently, check the documentation of the intrinsic
    /// to see how it's being used.
    ///
    /// Note that this means that an instance of `__m128i` typically just means
    /// a "bag of bits" which is left up to interpretation at the point of use.
    ///
    /// Most intrinsics using `__m128i` are prefixed with `_mm_` and the
    /// integer types tend to correspond to suffixes like "epi8" or "epi32".
    ///
    /// # Examples
    ///
    /// ```
    /// #[cfg(target_arch = "x86")]
    /// use std::arch::x86::*;
    /// #[cfg(target_arch = "x86_64")]
    /// use std::arch::x86_64::*;
    ///
    /// # fn main() {
    /// # #[target_feature(enable = "sse2")]
    /// # unsafe fn foo() {
    /// let all_bytes_zero = _mm_setzero_si128();
    /// let all_bytes_one = _mm_set1_epi8(1);
    /// let four_i32 = _mm_set_epi32(1, 2, 3, 4);
    /// # }
    /// # if is_x86_feature_detected!("sse2") { unsafe { foo() } }
    /// # }
    /// ```
    #[stable(feature = "simd_x86", since = "1.27.0")]
    pub struct __m128i(i64, i64);

    /// 128-bit wide set of four `f32` types, x86-specific
    ///
    /// This type is the same as the `__m128` type defined by Intel,
    /// representing a 128-bit SIMD register which internally is consisted of
    /// four packed `f32` instances. Usage of this type typically corresponds
    /// to the `sse` and up target features for x86/x86_64.
    ///
    /// Note that unlike `__m128i`, the integer version of the 128-bit
    /// registers, this `__m128` type has *one* interpretation. Each instance
    /// of `__m128` always corresponds to `f32x4`, or four `f32` types packed
    /// together.
    ///
    /// Most intrinsics using `__m128` are prefixed with `_mm_` and are
    /// suffixed with "ps" (or otherwise contain "ps"). Not to be confused with
    /// "pd" which is used for `__m128d`.
    ///
    /// # Examples
    ///
    /// ```
    /// #[cfg(target_arch = "x86")]
    /// use std::arch::x86::*;
    /// #[cfg(target_arch = "x86_64")]
    /// use std::arch::x86_64::*;
    ///
    /// # fn main() {
    /// # #[target_feature(enable = "sse")]
    /// # unsafe fn foo() {
    /// let four_zeros = _mm_setzero_ps();
    /// let four_ones = _mm_set1_ps(1.0);
    /// let four_floats = _mm_set_ps(1.0, 2.0, 3.0, 4.0);
    /// # }
    /// # if is_x86_feature_detected!("sse") { unsafe { foo() } }
    /// # }
    /// ```
    #[stable(feature = "simd_x86", since = "1.27.0")]
    pub struct __m128(f32, f32, f32, f32);

    /// 128-bit wide set of two `f64` types, x86-specific
    ///
    /// This type is the same as the `__m128d` type defined by Intel,
    /// representing a 128-bit SIMD register which internally is consisted of
    /// two packed `f64` instances. Usage of this type typically corresponds
    /// to the `sse` and up target features for x86/x86_64.
    ///
    /// Note that unlike `__m128i`, the integer version of the 128-bit
    /// registers, this `__m128d` type has *one* interpretation. Each instance
    /// of `__m128d` always corresponds to `f64x2`, or two `f64` types packed
    /// together.
    ///
    /// Most intrinsics using `__m128d` are prefixed with `_mm_` and are
    /// suffixed with "pd" (or otherwise contain "pd"). Not to be confused with
    /// "ps" which is used for `__m128`.
    ///
    /// # Examples
    ///
    /// ```
    /// #[cfg(target_arch = "x86")]
    /// use std::arch::x86::*;
    /// #[cfg(target_arch = "x86_64")]
    /// use std::arch::x86_64::*;
    ///
    /// # fn main() {
    /// # #[target_feature(enable = "sse")]
    /// # unsafe fn foo() {
    /// let two_zeros = _mm_setzero_pd();
    /// let two_ones = _mm_set1_pd(1.0);
    /// let two_floats = _mm_set_pd(1.0, 2.0);
    /// # }
    /// # if is_x86_feature_detected!("sse") { unsafe { foo() } }
    /// # }
    /// ```
    #[stable(feature = "simd_x86", since = "1.27.0")]
    pub struct __m128d(f64, f64);

    /// 256-bit wide integer vector type, x86-specific
    ///
    /// This type is the same as the `__m256i` type defined by Intel,
    /// representing a 256-bit SIMD register. Usage of this type typically
    /// corresponds to the `avx` and up target features for x86/x86_64.
    ///
    /// Internally this type may be viewed as:
    ///
    /// * `i8x32` - thirty two `i8` variables packed together
    /// * `i16x16` - sixteen `i16` variables packed together
    /// * `i32x8` - eight `i32` variables packed together
    /// * `i64x4` - four `i64` variables packed together
    ///
    /// (as well as unsigned versions). Each intrinsic may interpret the
    /// internal bits differently, check the documentation of the intrinsic
    /// to see how it's being used.
    ///
    /// Note that this means that an instance of `__m256i` typically just means
    /// a "bag of bits" which is left up to interpretation at the point of use.
    ///
    /// # Examples
    ///
    /// ```
    /// #[cfg(target_arch = "x86")]
    /// use std::arch::x86::*;
    /// #[cfg(target_arch = "x86_64")]
    /// use std::arch::x86_64::*;
    ///
    /// # fn main() {
    /// # #[target_feature(enable = "avx")]
    /// # unsafe fn foo() {
    /// let all_bytes_zero = _mm256_setzero_si256();
    /// let all_bytes_one = _mm256_set1_epi8(1);
    /// let eight_i32 = _mm256_set_epi32(1, 2, 3, 4, 5, 6, 7, 8);
    /// # }
    /// # if is_x86_feature_detected!("avx") { unsafe { foo() } }
    /// # }
    /// ```
    #[stable(feature = "simd_x86", since = "1.27.0")]
    pub struct __m256i(i64, i64, i64, i64);

    /// 256-bit wide set of eight `f32` types, x86-specific
    ///
    /// This type is the same as the `__m256` type defined by Intel,
    /// representing a 256-bit SIMD register which internally is consisted of
    /// eight packed `f32` instances. Usage of this type typically corresponds
    /// to the `avx` and up target features for x86/x86_64.
    ///
    /// Note that unlike `__m256i`, the integer version of the 256-bit
    /// registers, this `__m256` type has *one* interpretation. Each instance
    /// of `__m256` always corresponds to `f32x8`, or eight `f32` types packed
    /// together.
    ///
    /// Most intrinsics using `__m256` are prefixed with `_mm256_` and are
    /// suffixed with "ps" (or otherwise contain "ps"). Not to be confused with
    /// "pd" which is used for `__m256d`.
    ///
    /// # Examples
    ///
    /// ```
    /// #[cfg(target_arch = "x86")]
    /// use std::arch::x86::*;
    /// #[cfg(target_arch = "x86_64")]
    /// use std::arch::x86_64::*;
    ///
    /// # fn main() {
    /// # #[target_feature(enable = "avx")]
    /// # unsafe fn foo() {
    /// let eight_zeros = _mm256_setzero_ps();
    /// let eight_ones = _mm256_set1_ps(1.0);
    /// let eight_floats = _mm256_set_ps(1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0);
    /// # }
    /// # if is_x86_feature_detected!("avx") { unsafe { foo() } }
    /// # }
    /// ```
    #[stable(feature = "simd_x86", since = "1.27.0")]
    pub struct __m256(f32, f32, f32, f32, f32, f32, f32, f32);

    /// 256-bit wide set of four `f64` types, x86-specific
    ///
    /// This type is the same as the `__m256d` type defined by Intel,
    /// representing a 256-bit SIMD register which internally is consisted of
    /// four packed `f64` instances. Usage of this type typically corresponds
    /// to the `avx` and up target features for x86/x86_64.
    ///
    /// Note that unlike `__m256i`, the integer version of the 256-bit
    /// registers, this `__m256d` type has *one* interpretation. Each instance
    /// of `__m256d` always corresponds to `f64x4`, or four `f64` types packed
    /// together.
    ///
    /// Most intrinsics using `__m256d` are prefixed with `_mm256_` and are
    /// suffixed with "pd" (or otherwise contain "pd"). Not to be confused with
    /// "ps" which is used for `__m256`.
    ///
    /// # Examples
    ///
    /// ```
    /// #[cfg(target_arch = "x86")]
    /// use std::arch::x86::*;
    /// #[cfg(target_arch = "x86_64")]
    /// use std::arch::x86_64::*;
    ///
    /// # fn main() {
    /// # #[target_feature(enable = "avx")]
    /// # unsafe fn foo() {
    /// let four_zeros = _mm256_setzero_pd();
    /// let four_ones = _mm256_set1_pd(1.0);
    /// let four_floats = _mm256_set_pd(1.0, 2.0, 3.0, 4.0);
    /// # }
    /// # if is_x86_feature_detected!("avx") { unsafe { foo() } }
    /// # }
    /// ```
    #[stable(feature = "simd_x86", since = "1.27.0")]
    pub struct __m256d(f64, f64, f64, f64);

    /// 512-bit wide integer vector type, x86-specific
    ///
    /// This type is the same as the `__m512i` type defined by Intel,
    /// representing a 512-bit SIMD register. Usage of this type typically
    /// corresponds to the `avx512*` and up target features for x86/x86_64.
    ///
    /// Internally this type may be viewed as:
    ///
    /// * `i8x64` - sixty-four `i8` variables packed together
    /// * `i16x32` - thirty-two `i16` variables packed together
    /// * `i32x16` - sixteen `i32` variables packed together
    /// * `i64x8` - eight `i64` variables packed together
    ///
    /// (as well as unsigned versions). Each intrinsic may interpret the
    /// internal bits differently, check the documentation of the intrinsic
    /// to see how it's being used.
    ///
    /// Note that this means that an instance of `__m512i` typically just means
    /// a "bag of bits" which is left up to interpretation at the point of use.
    pub struct __m512i(i64, i64, i64, i64, i64, i64, i64, i64);

    /// 512-bit wide set of sixteen `f32` types, x86-specific
    ///
    /// This type is the same as the `__m512` type defined by Intel,
    /// representing a 512-bit SIMD register which internally is consisted of
    /// eight packed `f32` instances. Usage of this type typically corresponds
    /// to the `avx512*` and up target features for x86/x86_64.
    ///
    /// Note that unlike `__m512i`, the integer version of the 512-bit
    /// registers, this `__m512` type has *one* interpretation. Each instance
    /// of `__m512` always corresponds to `f32x16`, or sixteen `f32` types
    /// packed together.
    ///
    /// Most intrinsics using `__m512` are prefixed with `_mm512_` and are
    /// suffixed with "ps" (or otherwise contain "ps"). Not to be confused with
    /// "pd" which is used for `__m512d`.
    pub struct __m512(
        f32, f32, f32, f32, f32, f32, f32, f32,
        f32, f32, f32, f32, f32, f32, f32, f32,
    );

    /// 512-bit wide set of eight `f64` types, x86-specific
    ///
    /// This type is the same as the `__m512d` type defined by Intel,
    /// representing a 512-bit SIMD register which internally is consisted of
    /// eight packed `f64` instances. Usage of this type typically corresponds
    /// to the `avx` and up target features for x86/x86_64.
    ///
    /// Note that unlike `__m512i`, the integer version of the 512-bit
    /// registers, this `__m512d` type has *one* interpretation. Each instance
    /// of `__m512d` always corresponds to `f64x4`, or eight `f64` types packed
    /// together.
    ///
    /// Most intrinsics using `__m512d` are prefixed with `_mm512_` and are
    /// suffixed with "pd" (or otherwise contain "pd"). Not to be confused with
    /// "ps" which is used for `__m512`.
    pub struct __m512d(f64, f64, f64, f64, f64, f64, f64, f64);
}

/// The `__mmask16` type used in AVX-512 intrinsics, a 16-bit integer
#[allow(non_camel_case_types)]
pub type __mmask16 = i16;

#[cfg(test)]
mod test;
#[cfg(test)]
pub use self::test::*;

#[allow(non_camel_case_types)]
#[unstable(feature = "stdimd_internal", issue = "0")]
pub(crate) trait m128iExt: Sized {
    fn as_m128i(self) -> __m128i;

    #[inline]
    fn as_u8x16(self) -> crate::core_arch::simd::u8x16 {
        unsafe { transmute(self.as_m128i()) }
    }

    #[inline]
    fn as_u16x8(self) -> crate::core_arch::simd::u16x8 {
        unsafe { transmute(self.as_m128i()) }
    }

    #[inline]
    fn as_u32x4(self) -> crate::core_arch::simd::u32x4 {
        unsafe { transmute(self.as_m128i()) }
    }

    #[inline]
    fn as_u64x2(self) -> crate::core_arch::simd::u64x2 {
        unsafe { transmute(self.as_m128i()) }
    }

    #[inline]
    fn as_i8x16(self) -> crate::core_arch::simd::i8x16 {
        unsafe { transmute(self.as_m128i()) }
    }

    #[inline]
    fn as_i16x8(self) -> crate::core_arch::simd::i16x8 {
        unsafe { transmute(self.as_m128i()) }
    }

    #[inline]
    fn as_i32x4(self) -> crate::core_arch::simd::i32x4 {
        unsafe { transmute(self.as_m128i()) }
    }

    #[inline]
    fn as_i64x2(self) -> crate::core_arch::simd::i64x2 {
        unsafe { transmute(self.as_m128i()) }
    }
}

impl m128iExt for __m128i {
    #[inline]
    fn as_m128i(self) -> Self {
        self
    }
}

#[allow(non_camel_case_types)]
#[unstable(feature = "stdimd_internal", issue = "0")]
pub(crate) trait m256iExt: Sized {
    fn as_m256i(self) -> __m256i;

    #[inline]
    fn as_u8x32(self) -> crate::core_arch::simd::u8x32 {
        unsafe { transmute(self.as_m256i()) }
    }

    #[inline]
    fn as_u16x16(self) -> crate::core_arch::simd::u16x16 {
        unsafe { transmute(self.as_m256i()) }
    }

    #[inline]
    fn as_u32x8(self) -> crate::core_arch::simd::u32x8 {
        unsafe { transmute(self.as_m256i()) }
    }

    #[inline]
    fn as_u64x4(self) -> crate::core_arch::simd::u64x4 {
        unsafe { transmute(self.as_m256i()) }
    }

    #[inline]
    fn as_i8x32(self) -> crate::core_arch::simd::i8x32 {
        unsafe { transmute(self.as_m256i()) }
    }

    #[inline]
    fn as_i16x16(self) -> crate::core_arch::simd::i16x16 {
        unsafe { transmute(self.as_m256i()) }
    }

    #[inline]
    fn as_i32x8(self) -> crate::core_arch::simd::i32x8 {
        unsafe { transmute(self.as_m256i()) }
    }

    #[inline]
    fn as_i64x4(self) -> crate::core_arch::simd::i64x4 {
        unsafe { transmute(self.as_m256i()) }
    }
}

impl m256iExt for __m256i {
    #[inline]
    fn as_m256i(self) -> Self {
        self
    }
}

#[allow(non_camel_case_types)]
#[unstable(feature = "stdimd_internal", issue = "0")]
pub(crate) trait m512iExt: Sized {
    fn as_m512i(self) -> __m512i;

    #[inline]
    fn as_i32x16(self) -> crate::core_arch::simd::i32x16 {
        unsafe { transmute(self.as_m512i()) }
    }
}

impl m512iExt for __m512i {
    #[inline]
    fn as_m512i(self) -> Self {
        self
    }
}

mod eflags;
pub use self::eflags::*;

mod fxsr;
pub use self::fxsr::*;

mod bswap;
pub use self::bswap::*;

mod rdtsc;
pub use self::rdtsc::*;

mod cpuid;
pub use self::cpuid::*;
mod xsave;
pub use self::xsave::*;

mod sse;
pub use self::sse::*;
mod sse2;
pub use self::sse2::*;
mod sse3;
pub use self::sse3::*;
mod ssse3;
pub use self::ssse3::*;
mod sse41;
pub use self::sse41::*;
mod sse42;
pub use self::sse42::*;
mod avx;
pub use self::avx::*;
mod avx2;
pub use self::avx2::*;
mod fma;
pub use self::fma::*;

mod abm;
pub use self::abm::*;
mod bmi1;
pub use self::bmi1::*;

mod bmi2;
pub use self::bmi2::*;

#[cfg(not(stdarch_intel_sde))]
mod sse4a;
#[cfg(not(stdarch_intel_sde))]
pub use self::sse4a::*;

#[cfg(not(stdarch_intel_sde))]
mod tbm;
#[cfg(not(stdarch_intel_sde))]
pub use self::tbm::*;

mod mmx;
pub use self::mmx::*;

mod pclmulqdq;
pub use self::pclmulqdq::*;

mod aes;
pub use self::aes::*;

mod rdrand;
pub use self::rdrand::*;

mod sha;
pub use self::sha::*;

mod adx;
pub use self::adx::*;

#[cfg(test)]
use stdarch_test::assert_instr;

/// Generates the trap instruction `UD2`
#[cfg_attr(test, assert_instr(ud2))]
#[inline]
pub unsafe fn ud2() -> ! {
    intrinsics::abort()
}

mod avx512f;
pub use self::avx512f::*;

mod avx512ifma;
pub use self::avx512ifma::*;

mod bt;
pub use self::bt::*;

mod rtm;
pub use self::rtm::*;

mod f16c;
pub use self::f16c::*;