FM / core

/usr/local/lib64/python3.6/site-packages/torch/include/ATen/core/

/usr/local/lib64/python3.6/site-packages/torch/include/ATen/core

Name	Size	Mode	Actions
boxing/	-	0755	rm
dispatch/	-	0755	rm
op_registration/	-	0755	rm
alias_info.h	2986	0644	edit dl rm
Array.h	768	0644	edit dl rm
ATenGeneral.h	45	0644	edit dl rm
ATenOpList.h	246	0644	edit dl rm
aten_interned_strings.h	25389	0644	edit dl rm
Backtrace.h	59	0644	edit dl rm
blob.h	5422	0644	edit dl rm
builtin_function.h	3649	0644	edit dl rm
DeprecatedTypeProperties.h	3773	0644	edit dl rm
DeprecatedTypePropertiesRegistry.h	795	0644	edit dl rm
Dict.h	13195	0644	edit dl rm
Dict_inl.h	7996	0644	edit dl rm
Dimname.h	1188	0644	edit dl rm
DimVector.h	247	0644	edit dl rm
DistributionsHelper.h	12594	0644	edit dl rm
Formatting.h	959	0644	edit dl rm
function.h	2145	0644	edit dl rm
functional.h	1460	0644	edit dl rm
function_schema.h	13577	0644	edit dl rm
function_schema_inl.h	9319	0644	edit dl rm
Generator.h	4935	0644	edit dl rm
grad_mode.h	210	0644	edit dl rm
interned_strings.h	25332	0644	edit dl rm
interned_strings_class.h	770	0644	edit dl rm
ivalue.h	38823	0644	edit dl rm
ivalue_inl.h	59963	0644	edit dl rm
ivalue_to.h	756	0644	edit dl rm
jit_type.h	75971	0644	edit dl rm
jit_type_base.h	6508	0644	edit dl rm
LegacyTypeDispatch.h	4626	0644	edit dl rm
List.h	15667	0644	edit dl rm
List_inl.h	11012	0644	edit dl rm
Macros.h	44	0644	edit dl rm
MT19937RNGEngine.h	6410	0644	edit dl rm
NamedTensor.h	5050	0644	edit dl rm
operator_name.h	3018	0644	edit dl rm
PhiloxRNGEngine.h	6496	0644	edit dl rm
PythonModeTLS.h	403	0644	edit dl rm
qualified_name.h	4358	0644	edit dl rm
QuantizerBase.h	2443	0644	edit dl rm
Range.h	418	0644	edit dl rm
Reduction.h	461	0644	edit dl rm
rref_interface.h	1144	0644	edit dl rm
Scalar.h	29	0644	edit dl rm
ScalarType.h	33	0644	edit dl rm
stack.h	6034	0644	edit dl rm
Tensor.h	1756	0644	edit dl rm
TensorAccessor.h	10296	0644	edit dl rm
TensorBase.h	32767	0644	edit dl rm
TensorBody.h	247555	0644	edit dl rm
TransformationHelper.h	6911	0644	edit dl rm
typeid.h	29	0644	edit dl rm
UndefinedTensorImpl.h	42	0644	edit dl rm
UnsafeFromTH.h	708	0644	edit dl rm
VariableHooksInterface.h	3312	0644	edit dl rm
Variadic.h	2257	0644	edit dl rm
Vitals.h	2305	0644	edit dl rm

Edit: /usr/local/lib64/python3.6/site-packages/torch/include/ATen/core/MT19937RNGEngine.h (6410B)

#pragma once

// define constants like M_PI and C keywords for MSVC
#ifdef _MSC_VER
#ifndef _USE_MATH_DEFINES
#define _USE_MATH_DEFINES
#endif
#include <math.h>
#endif

#include <stdint.h>
#include <cmath>
#include <array>

namespace at {

constexpr int MERSENNE_STATE_N = 624;
constexpr int MERSENNE_STATE_M = 397;
constexpr uint32_t MATRIX_A = 0x9908b0df;
constexpr uint32_t UMASK = 0x80000000;
constexpr uint32_t LMASK = 0x7fffffff;

/**
 * Note [Mt19937 Engine implementation]
 * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 * Originally implemented in:
 * http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/MT2002/CODES/MTARCOK/mt19937ar-cok.c
 * and modified with C++ constructs. Moreover the state array of the engine
 * has been modified to hold 32 bit uints instead of 64 bits.
 *
 * Note that we reimplemented mt19937 instead of using std::mt19937 because,
 * at::mt19937 turns out to be faster in the pytorch codebase. PyTorch builds with -O2
 * by default and following are the benchmark numbers (benchmark code can be found at
 * https://github.com/syed-ahmed/benchmark-rngs):
 *
 * with -O2
 * Time to get 100000000 philox randoms with at::uniform_real_distribution = 0.462759s
 * Time to get 100000000 at::mt19937 randoms with at::uniform_real_distribution = 0.39628s
 * Time to get 100000000 std::mt19937 randoms with std::uniform_real_distribution = 0.352087s
 * Time to get 100000000 std::mt19937 randoms with at::uniform_real_distribution = 0.419454s
 *
 * std::mt19937 is faster when used in conjunction with std::uniform_real_distribution,
 * however we can't use std::uniform_real_distribution because of this bug:
 * http://open-std.org/JTC1/SC22/WG21/docs/lwg-active.html#2524. Plus, even if we used
 * std::uniform_real_distribution and filtered out the 1's, it is a different algorithm
 * than what's in pytorch currently and that messes up the tests in tests_distributions.py.
 * The other option, using std::mt19937 with at::uniform_real_distribution is a tad bit slower
 * than at::mt19937 with at::uniform_real_distribution and hence, we went with the latter.
 *
 * Copyright notice:
 * A C-program for MT19937, with initialization improved 2002/2/10.
 * Coded by Takuji Nishimura and Makoto Matsumoto.
 * This is a faster version by taking Shawn Cokus's optimization,
 * Matthe Bellew's simplification, Isaku Wada's real version.
 *
 * Before using, initialize the state by using init_genrand(seed)
 * or init_by_array(init_key, key_length).
 *
 * Copyright (C) 1997 - 2002, Makoto Matsumoto and Takuji Nishimura,
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 *
 *   1. Redistributions of source code must retain the above copyright
 *   notice, this list of conditions and the following disclaimer.
 *
 *   2. Redistributions in binary form must reproduce the above copyright
 *   notice, this list of conditions and the following disclaimer in the
 *   documentation and/or other materials provided with the distribution.
 *
 *   3. The names of its contributors may not be used to endorse or promote
 *   products derived from this software without specific prior written
 *   permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 *
 * Any feedback is very welcome.
 * http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html
 * email: m-mat @ math.sci.hiroshima-u.ac.jp (remove space)
 */

/**
 * mt19937_data_pod is used to get POD data in and out
 * of mt19937_engine. Used in torch.get_rng_state and
 * torch.set_rng_state functions.
 */
struct mt19937_data_pod {
  uint64_t seed_;
  int left_;
  bool seeded_;
  uint32_t next_;
  std::array<uint32_t, MERSENNE_STATE_N> state_;
};

class mt19937_engine {
public:

inline explicit mt19937_engine(uint64_t seed = 5489) {
    init_with_uint32(seed);
  }

inline mt19937_data_pod data() const {
    return data_;
  }

inline void set_data(mt19937_data_pod data) {
    data_ = data;
  }

inline uint64_t seed() const {
    return data_.seed_;
  }

inline bool is_valid() {
    if ((data_.seeded_ == true)
      && (data_.left_ > 0 && data_.left_ <= MERSENNE_STATE_N)
      && (data_.next_ <= MERSENNE_STATE_N)) {
      return true;
    }
    return false;
  }

inline uint32_t operator()() {
    uint32_t y;

if (--(data_.left_) == 0) {
        next_state();
    }
    y = *(data_.state_.data() + data_.next_++);
    y ^= (y >> 11);
    y ^= (y << 7) & 0x9d2c5680;
    y ^= (y << 15) & 0xefc60000;
    y ^= (y >> 18);

return y;
  }

private:
  mt19937_data_pod data_;

inline void init_with_uint32(uint64_t seed) {
    data_.seed_ = seed;
    data_.seeded_ = true;
    data_.state_[0] = seed & 0xffffffff;
    for(int j = 1; j < MERSENNE_STATE_N; j++) {
      data_.state_[j] = (1812433253 * (data_.state_[j-1] ^ (data_.state_[j-1] >> 30)) + j);
    }
    data_.left_ = 1;
    data_.next_ = 0;
  }

inline uint32_t mix_bits(uint32_t u, uint32_t v) {
    return (u & UMASK) | (v & LMASK);
  }

inline uint32_t twist(uint32_t u, uint32_t v) {
    return (mix_bits(u,v) >> 1) ^ (v & 1 ? MATRIX_A : 0);
  }

inline void next_state() {
    uint32_t* p = data_.state_.data();
    data_.left_ = MERSENNE_STATE_N;
    data_.next_ = 0;

for(int j = MERSENNE_STATE_N - MERSENNE_STATE_M + 1; --j; p++) {
      *p = p[MERSENNE_STATE_M] ^ twist(p[0], p[1]);
    }

for(int j = MERSENNE_STATE_M; --j; p++) {
      *p = p[MERSENNE_STATE_M - MERSENNE_STATE_N] ^ twist(p[0], p[1]);
    }

*p = p[MERSENNE_STATE_M - MERSENNE_STATE_N] ^ twist(p[0], data_.state_[0]);
  }

};

typedef mt19937_engine mt19937;

} // namespace at

cmd: