cpu_info.h 1.9 KB
Newer Older
1
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved.
L
liaogang 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */

#pragma once

#include <stddef.h>

P
peizhilin 已提交
19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38
#ifdef _WIN32
#if defined(__AVX2__)
#include <immintrin.h>  //avx2
#elif defined(__AVX__)
#include <intrin.h>  //avx
#endif               // AVX
#else                // WIN32
#ifdef __AVX__
#include <immintrin.h>
#endif
#endif  // WIN32

#if defined(_WIN32)
#define ALIGN32_BEG __declspec(align(32))
#define ALIGN32_END
#else
#define ALIGN32_BEG
#define ALIGN32_END __attribute__((aligned(32)))
#endif  // _WIN32

L
liaogang 已提交
39 40 41
namespace paddle {
namespace platform {

S
sneaxiy 已提交
42 43
size_t CpuTotalPhysicalMemory();

L
liaogang 已提交
44 45 46
//! Get the maximum allocation size for a machine.
size_t CpuMaxAllocSize();

47 48 49
//! Get the maximum allocation size for a machine.
size_t CUDAPinnedMaxAllocSize();

L
liaogang 已提交
50 51 52 53 54
//! Get the minimum chunk size for buddy allocator.
size_t CpuMinChunkSize();

//! Get the maximum chunk size for buddy allocator.
size_t CpuMaxChunkSize();
L
liaogang 已提交
55

56 57 58 59 60 61
//! Get the minimum chunk size for buddy allocator.
size_t CUDAPinnedMinChunkSize();

//! Get the maximum chunk size for buddy allocator.
size_t CUDAPinnedMaxChunkSize();

T
tensor-tang 已提交
62 63 64
typedef enum {
  isa_any,
  sse42,
T
tensor-tang 已提交
65
  avx,
T
tensor-tang 已提交
66
  avx2,
T
tensor-tang 已提交
67
  avx512f,
T
tensor-tang 已提交
68 69 70 71 72 73 74
  avx512_core,
  avx512_core_vnni,
  avx512_mic,
  avx512_mic_4ops,
} cpu_isa_t;  // Instruction set architecture

// May I use some instruction
T
tensor-tang 已提交
75
bool MayIUse(const cpu_isa_t cpu_isa);
T
tensor-tang 已提交
76

L
liaogang 已提交
77 78
}  // namespace platform
}  // namespace paddle