mirror of
https://github.com/boostorg/log.git
synced 2026-02-11 11:52:20 +00:00
351 lines
14 KiB
C++
351 lines
14 KiB
C++
/*
|
|
* Copyright Andrey Semashev 2007 - 2014.
|
|
* Distributed under the Boost Software License, Version 1.0.
|
|
* (See accompanying file LICENSE_1_0.txt or copy at
|
|
* http://www.boost.org/LICENSE_1_0.txt)
|
|
*/
|
|
/*!
|
|
* \file dump_avx2.cpp
|
|
* \author Andrey Semashev
|
|
* \date 05.05.2013
|
|
*
|
|
* \brief This header is the Boost.Log library implementation, see the library documentation
|
|
* at http://www.boost.org/doc/libs/release/libs/log/doc/html/index.html.
|
|
*/
|
|
|
|
// NOTE: You should generally avoid including headers as much as possible here, because this file
|
|
// is compiled with special compiler options, and any included header may result in generation of
|
|
// unintended code with these options and violation of ODR.
|
|
#include <ostream>
|
|
#include <immintrin.h>
|
|
#include <boost/cstdint.hpp>
|
|
#include <boost/log/detail/config.hpp>
|
|
#include <boost/log/detail/header.hpp>
|
|
|
|
#if defined(__x86_64) || defined(__x86_64__) || \
|
|
defined(__amd64__) || defined(__amd64) || \
|
|
defined(_M_X64)
|
|
#define BOOST_LOG_AUX_X86_64
|
|
#endif
|
|
|
|
namespace boost {
|
|
|
|
BOOST_LOG_OPEN_NAMESPACE
|
|
|
|
namespace aux {
|
|
|
|
extern const char g_lowercase_dump_char_table[];
|
|
extern const char g_uppercase_dump_char_table[];
|
|
|
|
template< typename CharT >
|
|
extern void dump_data_generic(const void* data, std::size_t size, std::basic_ostream< CharT >& strm);
|
|
|
|
BOOST_LOG_ANONYMOUS_NAMESPACE {
|
|
|
|
enum
|
|
{
|
|
packs_per_stride = 32,
|
|
stride = packs_per_stride * 32
|
|
};
|
|
|
|
union ymm_constant
|
|
{
|
|
uint8_t as_bytes[32];
|
|
__m256i as_mm;
|
|
|
|
BOOST_FORCEINLINE operator __m256i () const { return as_mm; }
|
|
};
|
|
|
|
static const ymm_constant mm_shuffle_pattern1 = {{ 0x80, 0, 1, 0x80, 2, 3, 0x80, 4, 5, 0x80, 6, 7, 0x80, 8, 9, 0x80, 0x80, 0, 1, 0x80, 2, 3, 0x80, 4, 5, 0x80, 6, 7, 0x80, 8, 9, 0x80 }};
|
|
static const ymm_constant mm_shuffle_pattern2 = {{ 0, 1, 0x80, 2, 3, 0x80, 4, 5, 0x80, 6, 7, 0x80, 8, 9, 0x80, 10, 0, 1, 0x80, 2, 3, 0x80, 4, 5, 0x80, 6, 7, 0x80, 8, 9, 0x80, 10 }};
|
|
static const ymm_constant mm_shuffle_pattern3 = {{ 5, 0x80, 6, 7, 0x80, 8, 9, 0x80, 10, 11, 0x80, 12, 13, 0x80, 14, 15, 5, 0x80, 6, 7, 0x80, 8, 9, 0x80, 10, 11, 0x80, 12, 13, 0x80, 14, 15 }};
|
|
static const ymm_constant mm_shuffle_pattern13 = {{ 0x80, 0, 1, 0x80, 2, 3, 0x80, 4, 5, 0x80, 6, 7, 0x80, 8, 9, 0x80, 5, 0x80, 6, 7, 0x80, 8, 9, 0x80, 10, 11, 0x80, 12, 13, 0x80, 14, 15 }};
|
|
|
|
#if defined(BOOST_LOG_AUX_X86_64)
|
|
|
|
// x86-64 architecture has more registers which we can utilize to pass constants
|
|
#define BOOST_LOG_AUX_MM_CONSTANT_ARGS_DECL __m256i mm_15, __m256i mm_9, __m256i mm_char_0, __m256i mm_char_space,
|
|
#define BOOST_LOG_AUX_MM_CONSTANT_ARGS mm_15, mm_9, mm_char_0, mm_char_space,
|
|
#define BOOST_LOG_AUX_MM_CONSTANTS \
|
|
const __m256i mm_15 = _mm256_set1_epi32(0x0F0F0F0F);\
|
|
const __m256i mm_9 = _mm256_set1_epi32(0x09090909);\
|
|
const __m256i mm_char_0 = _mm256_set1_epi32(0x30303030);\
|
|
const __m256i mm_char_space = _mm256_set1_epi32(0x20202020);
|
|
|
|
#else
|
|
|
|
// MSVC in 32-bit mode is not able to pass all constants to dump_pack, and is also not able to align them on the stack, so we have to fetch them from global constants
|
|
static const ymm_constant mm_15 = {{ 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F, 0x0F }};
|
|
static const ymm_constant mm_9 = {{ 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09, 0x09 }};
|
|
static const ymm_constant mm_char_0 = {{ 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30 }};
|
|
static const ymm_constant mm_char_space = {{ 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20 }};
|
|
#define BOOST_LOG_AUX_MM_CONSTANT_ARGS_DECL
|
|
#define BOOST_LOG_AUX_MM_CONSTANT_ARGS
|
|
#define BOOST_LOG_AUX_MM_CONSTANTS
|
|
|
|
#endif
|
|
|
|
/*!
|
|
* \brief Dumps a pack of input data into a string of 8 bit ASCII characters.
|
|
*
|
|
* The composed string is placed as follows (in Intel notation): mm_output1[127:0], mm_output2[127:0], mm_output3[127:0], mm_output1[255:128], mm_output2[255:128], mm_output3[255:128].
|
|
*/
|
|
static BOOST_FORCEINLINE void dump_pack
|
|
(
|
|
BOOST_LOG_AUX_MM_CONSTANT_ARGS_DECL
|
|
__m256i mm_char_10_to_a, __m256i mm_input,
|
|
__m256i& mm_output1, __m256i& mm_output2, __m256i& mm_output3
|
|
)
|
|
{
|
|
// Split half-bytes
|
|
__m256i mm_input_hi = _mm256_and_si256(_mm256_srli_epi16(mm_input, 4), mm_15);
|
|
__m256i mm_input_lo = _mm256_and_si256(mm_input, mm_15);
|
|
|
|
// Stringize each of the halves
|
|
__m256i mm_addend_hi = _mm256_cmpgt_epi8(mm_input_hi, mm_9);
|
|
__m256i mm_addend_lo = _mm256_cmpgt_epi8(mm_input_lo, mm_9);
|
|
mm_addend_hi = _mm256_and_si256(mm_char_10_to_a, mm_addend_hi);
|
|
mm_addend_lo = _mm256_and_si256(mm_char_10_to_a, mm_addend_lo);
|
|
|
|
mm_input_hi = _mm256_add_epi8(mm_input_hi, mm_char_0);
|
|
mm_input_lo = _mm256_add_epi8(mm_input_lo, mm_char_0);
|
|
|
|
mm_input_hi = _mm256_add_epi8(mm_input_hi, mm_addend_hi);
|
|
mm_input_lo = _mm256_add_epi8(mm_input_lo, mm_addend_lo);
|
|
|
|
// Join them back together
|
|
__m256i mm_1 = _mm256_unpacklo_epi8(mm_input_hi, mm_input_lo);
|
|
__m256i mm_2 = _mm256_unpackhi_epi8(mm_input_hi, mm_input_lo);
|
|
|
|
// Insert spaces between stringized bytes:
|
|
// |0123456789abcdef|0123456789abcdef|
|
|
// | 01 23 45 67 89 |ab cd ef 01 23 4|5 67 89 ab cd ef|
|
|
__m256i mm_out1 = _mm256_shuffle_epi8(mm_1, mm_shuffle_pattern1.as_mm);
|
|
__m256i mm_out3 = _mm256_shuffle_epi8(mm_2, mm_shuffle_pattern3.as_mm);
|
|
__m256i mm_out2 = _mm256_shuffle_epi8(_mm256_alignr_epi8(mm_2, mm_1, 10), mm_shuffle_pattern2.as_mm);
|
|
|
|
mm_output1 = _mm256_max_epu8(mm_out1, mm_char_space);
|
|
mm_output2 = _mm256_max_epu8(mm_out2, mm_char_space);
|
|
mm_output3 = _mm256_max_epu8(mm_out3, mm_char_space);
|
|
}
|
|
|
|
//! Dumps a pack of input data into a string of 8 bit ASCII characters
|
|
static BOOST_FORCEINLINE void dump_pack
|
|
(
|
|
BOOST_LOG_AUX_MM_CONSTANT_ARGS_DECL
|
|
__m256i mm_char_10_to_a, __m128i mm_input,
|
|
__m128i& mm_output1, __m128i& mm_output2, __m128i& mm_output3
|
|
)
|
|
{
|
|
// Split half-bytes
|
|
__m128i mm_input_hi = _mm_srli_epi16(mm_input, 4);
|
|
__m256i mm = _mm256_insertf128_si256(_mm256_castsi128_si256(_mm_unpacklo_epi8(mm_input_hi, mm_input)), _mm_unpackhi_epi8(mm_input_hi, mm_input), 1);
|
|
mm = _mm256_and_si256(mm, mm_15);
|
|
|
|
// Stringize the halves
|
|
__m256i mm_addend = _mm256_cmpgt_epi8(mm, mm_9);
|
|
mm_addend = _mm256_and_si256(mm_char_10_to_a, mm_addend);
|
|
|
|
mm = _mm256_add_epi8(mm, mm_char_0);
|
|
mm = _mm256_add_epi8(mm, mm_addend);
|
|
|
|
// Insert spaces between stringized bytes:
|
|
__m256i mm_out13 = _mm256_shuffle_epi8(mm, mm_shuffle_pattern13.as_mm);
|
|
__m128i mm_out2 = _mm_shuffle_epi8(_mm_alignr_epi8(_mm256_extractf128_si256(mm, 1), _mm256_castsi256_si128(mm), 10), _mm256_castsi256_si128(mm_shuffle_pattern2.as_mm));
|
|
|
|
mm_out13 = _mm256_max_epu8(mm_out13, mm_char_space);
|
|
mm_output2 = _mm_max_epu8(mm_out2, _mm256_castsi256_si128(mm_char_space));
|
|
mm_output1 = _mm256_castsi256_si128(mm_out13);
|
|
mm_output3 = _mm256_extractf128_si256(mm_out13, 1);
|
|
}
|
|
|
|
template< typename CharT >
|
|
BOOST_FORCEINLINE void store_characters(__m128i mm_chars, CharT* buf)
|
|
{
|
|
switch (sizeof(CharT))
|
|
{
|
|
case 1:
|
|
_mm_store_si128(reinterpret_cast< __m128i* >(buf), mm_chars);
|
|
break;
|
|
|
|
case 2:
|
|
_mm256_store_si256(reinterpret_cast< __m256i* >(buf), _mm256_cvtepu8_epi16(mm_chars));
|
|
break;
|
|
|
|
case 4:
|
|
{
|
|
__m128i mm = _mm_unpackhi_epi64(mm_chars, mm_chars);
|
|
_mm256_store_si256(reinterpret_cast< __m256i* >(buf), _mm256_cvtepu8_epi32(mm_chars));
|
|
_mm256_store_si256(reinterpret_cast< __m256i* >(buf) + 1, _mm256_cvtepu8_epi32(mm));
|
|
}
|
|
break;
|
|
}
|
|
}
|
|
|
|
template< typename CharT >
|
|
BOOST_FORCEINLINE void store_characters_x3(__m256i mm_chars1, __m256i mm_chars2, __m256i mm_chars3, CharT* buf)
|
|
{
|
|
store_characters(_mm256_castsi256_si128(mm_chars1), buf);
|
|
store_characters(_mm256_castsi256_si128(mm_chars2), buf + 16);
|
|
store_characters(_mm256_castsi256_si128(mm_chars3), buf + 32);
|
|
store_characters(_mm256_extracti128_si256(mm_chars1, 1), buf + 48);
|
|
store_characters(_mm256_extracti128_si256(mm_chars2, 1), buf + 64);
|
|
store_characters(_mm256_extracti128_si256(mm_chars3, 1), buf + 80);
|
|
}
|
|
|
|
template< typename CharT >
|
|
BOOST_FORCEINLINE void dump_data_avx2(const void* data, std::size_t size, std::basic_ostream< CharT >& strm)
|
|
{
|
|
typedef CharT char_type;
|
|
|
|
char_type buf_storage[stride * 3u + 32u];
|
|
// Align the temporary buffer at 32 bytes
|
|
char_type* const buf = reinterpret_cast< char_type* >((uint8_t*)buf_storage + (32u - (((uintptr_t)(char_type*)buf_storage) & 31u)));
|
|
char_type* buf_begin = buf + 1u; // skip the first space of the first chunk
|
|
char_type* buf_end = buf + stride * 3u;
|
|
|
|
__m256i mm_char_10_to_a;
|
|
if (strm.flags() & std::ios_base::uppercase)
|
|
mm_char_10_to_a = _mm256_set1_epi32(0x07070707); // '9' is 0x39 and 'A' is 0x41 in ASCII, so we have to add 0x07 to 0x3A to get uppercase letters
|
|
else
|
|
mm_char_10_to_a = _mm256_set1_epi32(0x27272727); // ...and 'a' is 0x61, which means we have to add 0x27 to 0x3A to get lowercase letters
|
|
|
|
// First, check the input alignment. Also, if we can dump the whole data in one go, do it right away. It turns out to be faster than splitting
|
|
// the work between prealign and tail part. It is also a fairly common case since on most platforms memory is not aligned to 32 bytes (i.e. prealign is often needed).
|
|
const uint8_t* p = static_cast< const uint8_t* >(data);
|
|
const std::size_t prealign_size = size == 32u ? static_cast< std::size_t >(32u) : static_cast< std::size_t >((32u - ((uintptr_t)p & 31u)) & 31u);
|
|
if (prealign_size)
|
|
{
|
|
__m256i mm_input = _mm256_lddqu_si256(reinterpret_cast< const __m256i* >(p));
|
|
BOOST_LOG_AUX_MM_CONSTANTS
|
|
|
|
__m256i mm_output1, mm_output2, mm_output3;
|
|
dump_pack(BOOST_LOG_AUX_MM_CONSTANT_ARGS mm_char_10_to_a, mm_input, mm_output1, mm_output2, mm_output3);
|
|
|
|
store_characters_x3(mm_output1, mm_output2, mm_output3, buf);
|
|
|
|
_mm256_zeroall(); // need to zero all ymm registers to avoid register spills/restores the compler generates around the function call
|
|
strm.write(buf_begin, prealign_size * 3u - 1u);
|
|
|
|
buf_begin = buf;
|
|
size -= prealign_size;
|
|
p += prealign_size;
|
|
}
|
|
|
|
const std::size_t stride_count = size / stride;
|
|
std::size_t tail_size = size % stride;
|
|
for (std::size_t i = 0; i < stride_count; ++i)
|
|
{
|
|
char_type* b = buf;
|
|
BOOST_LOG_AUX_MM_CONSTANTS
|
|
|
|
for (unsigned int j = 0; j < packs_per_stride; ++j, b += 3u * 32u, p += 32u)
|
|
{
|
|
__m256i mm_input = _mm256_load_si256(reinterpret_cast< const __m256i* >(p));
|
|
__m256i mm_output1, mm_output2, mm_output3;
|
|
dump_pack(BOOST_LOG_AUX_MM_CONSTANT_ARGS mm_char_10_to_a, mm_input, mm_output1, mm_output2, mm_output3);
|
|
|
|
store_characters_x3(mm_output1, mm_output2, mm_output3, b);
|
|
}
|
|
|
|
_mm256_zeroall(); // need to zero all ymm registers to avoid register spills/restores the compler generates around the function call
|
|
strm.write(buf_begin, buf_end - buf_begin);
|
|
buf_begin = buf;
|
|
}
|
|
|
|
if (BOOST_UNLIKELY(tail_size > 0))
|
|
{
|
|
char_type* b = buf;
|
|
while (tail_size >= 16u)
|
|
{
|
|
__m128i mm_input = _mm_load_si128(reinterpret_cast< const __m128i* >(p));
|
|
BOOST_LOG_AUX_MM_CONSTANTS
|
|
|
|
__m128i mm_output1, mm_output2, mm_output3;
|
|
dump_pack(BOOST_LOG_AUX_MM_CONSTANT_ARGS mm_char_10_to_a, mm_input, mm_output1, mm_output2, mm_output3);
|
|
|
|
store_characters(mm_output1, b);
|
|
store_characters(mm_output2, b + 16u);
|
|
store_characters(mm_output3, b + 32u);
|
|
|
|
b += 3u * 16u;
|
|
p += 16u;
|
|
tail_size -= 16u;
|
|
}
|
|
|
|
_mm256_zeroall(); // need to zero all ymm registers to avoid register spills/restores the compler generates around the function call
|
|
const char* const char_table = (strm.flags() & std::ios_base::uppercase) ? g_uppercase_dump_char_table : g_lowercase_dump_char_table;
|
|
for (unsigned int i = 0; i < tail_size; ++i, ++p, b += 3u)
|
|
{
|
|
uint32_t n = *p;
|
|
b[0] = static_cast< char_type >(' ');
|
|
b[1] = static_cast< char_type >(char_table[n >> 4]);
|
|
b[2] = static_cast< char_type >(char_table[n & 0x0F]);
|
|
}
|
|
|
|
strm.write(buf_begin, b - buf_begin);
|
|
}
|
|
}
|
|
|
|
} // namespace
|
|
|
|
void dump_data_char_avx2(const void* data, std::size_t size, std::basic_ostream< char >& strm)
|
|
{
|
|
if (size >= 32)
|
|
{
|
|
dump_data_avx2(data, size, strm);
|
|
}
|
|
else
|
|
{
|
|
dump_data_generic(data, size, strm);
|
|
}
|
|
}
|
|
|
|
void dump_data_wchar_avx2(const void* data, std::size_t size, std::basic_ostream< wchar_t >& strm)
|
|
{
|
|
if (size >= 32)
|
|
{
|
|
dump_data_avx2(data, size, strm);
|
|
}
|
|
else
|
|
{
|
|
dump_data_generic(data, size, strm);
|
|
}
|
|
}
|
|
|
|
#if !defined(BOOST_NO_CXX11_CHAR16_T)
|
|
void dump_data_char16_avx2(const void* data, std::size_t size, std::basic_ostream< char16_t >& strm)
|
|
{
|
|
if (size >= 32)
|
|
{
|
|
dump_data_avx2(data, size, strm);
|
|
}
|
|
else
|
|
{
|
|
dump_data_generic(data, size, strm);
|
|
}
|
|
}
|
|
#endif
|
|
|
|
#if !defined(BOOST_NO_CXX11_CHAR32_T)
|
|
void dump_data_char32_avx2(const void* data, std::size_t size, std::basic_ostream< char32_t >& strm)
|
|
{
|
|
if (size >= 32)
|
|
{
|
|
dump_data_avx2(data, size, strm);
|
|
}
|
|
else
|
|
{
|
|
dump_data_generic(data, size, strm);
|
|
}
|
|
}
|
|
#endif
|
|
|
|
} // namespace aux
|
|
|
|
BOOST_LOG_CLOSE_NAMESPACE // namespace log
|
|
|
|
} // namespace boost
|
|
|
|
#include <boost/log/detail/footer.hpp>
|