##// END OF EJS Templates
perf: add command to benchmark bundle reading...
perf: add command to benchmark bundle reading Upcoming commits will be refactoring bundle2 I/O code. This commit establishes a `hg perfbundleread` command that measures how long it takes to read a bundle using various mechanisms. As a baseline, here's output from an uncompressed bundle1 bundle of my Firefox repo (7,098,622,890 bytes): ! read(8k) ! wall 0.763481 comb 0.760000 user 0.160000 sys 0.600000 (best of 6) ! read(16k) ! wall 0.644512 comb 0.640000 user 0.110000 sys 0.530000 (best of 16) ! read(32k) ! wall 0.581172 comb 0.590000 user 0.060000 sys 0.530000 (best of 18) ! read(128k) ! wall 0.535183 comb 0.530000 user 0.010000 sys 0.520000 (best of 19) ! cg1 deltaiter() ! wall 0.873500 comb 0.880000 user 0.840000 sys 0.040000 (best of 12) ! cg1 getchunks() ! wall 6.283797 comb 6.270000 user 5.570000 sys 0.700000 (best of 3) ! cg1 read(8k) ! wall 1.097173 comb 1.100000 user 0.400000 sys 0.700000 (best of 10) ! cg1 read(16k) ! wall 0.810750 comb 0.800000 user 0.200000 sys 0.600000 (best of 13) ! cg1 read(32k) ! wall 0.671215 comb 0.670000 user 0.110000 sys 0.560000 (best of 15) ! cg1 read(128k) ! wall 0.597857 comb 0.600000 user 0.020000 sys 0.580000 (best of 15) And from an uncompressed bundle2 bundle (6,070,036,163 bytes): ! read(8k) ! wall 0.676997 comb 0.680000 user 0.160000 sys 0.520000 (best of 15) ! read(16k) ! wall 0.592706 comb 0.590000 user 0.080000 sys 0.510000 (best of 17) ! read(32k) ! wall 0.529395 comb 0.530000 user 0.050000 sys 0.480000 (best of 16) ! read(128k) ! wall 0.491270 comb 0.490000 user 0.010000 sys 0.480000 (best of 19) ! bundle2 forwardchunks() ! wall 2.997131 comb 2.990000 user 2.270000 sys 0.720000 (best of 4) ! bundle2 iterparts() ! wall 12.247197 comb 10.670000 user 8.170000 sys 2.500000 (best of 3) ! bundle2 part seek() ! wall 11.761675 comb 10.500000 user 8.240000 sys 2.260000 (best of 3) ! bundle2 part read(8k) ! wall 9.116163 comb 9.110000 user 8.240000 sys 0.870000 (best of 3) ! bundle2 part read(16k) ! wall 8.984362 comb 8.970000 user 8.110000 sys 0.860000 (best of 3) ! bundle2 part read(32k) ! wall 8.758364 comb 8.740000 user 7.860000 sys 0.880000 (best of 3) ! bundle2 part read(128k) ! wall 8.749040 comb 8.730000 user 7.830000 sys 0.900000 (best of 3) We already see some interesting data. Notably that bundle2 has significant overhead compared to bundle1. This matters for e.g. stream clone bundles, which can be applied at >1Gbps. Differential Revision: https://phab.mercurial-scm.org/D1385

File last commit:

r30895:c32454d6 default
r35108:e9661304 default
Show More
mem.h
372 lines | 11.3 KiB | text/x-c | CLexer
/**
* Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
* All rights reserved.
*
* This source code is licensed under the BSD-style license found in the
* LICENSE file in the root directory of this source tree. An additional grant
* of patent rights can be found in the PATENTS file in the same directory.
*/
#ifndef MEM_H_MODULE
#define MEM_H_MODULE
#if defined (__cplusplus)
extern "C" {
#endif
/*-****************************************
* Dependencies
******************************************/
#include <stddef.h> /* size_t, ptrdiff_t */
#include <string.h> /* memcpy */
/*-****************************************
* Compiler specifics
******************************************/
#if defined(_MSC_VER) /* Visual Studio */
# include <stdlib.h> /* _byteswap_ulong */
# include <intrin.h> /* _byteswap_* */
#endif
#if defined(__GNUC__)
# define MEM_STATIC static __inline __attribute__((unused))
#elif defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
# define MEM_STATIC static inline
#elif defined(_MSC_VER)
# define MEM_STATIC static __inline
#else
# define MEM_STATIC static /* this version may generate warnings for unused static functions; disable the relevant warning */
#endif
/* code only tested on 32 and 64 bits systems */
#define MEM_STATIC_ASSERT(c) { enum { MEM_static_assert = 1/(int)(!!(c)) }; }
MEM_STATIC void MEM_check(void) { MEM_STATIC_ASSERT((sizeof(size_t)==4) || (sizeof(size_t)==8)); }
/*-**************************************************************
* Basic Types
*****************************************************************/
#if !defined (__VMS) && (defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
# include <stdint.h>
typedef uint8_t BYTE;
typedef uint16_t U16;
typedef int16_t S16;
typedef uint32_t U32;
typedef int32_t S32;
typedef uint64_t U64;
typedef int64_t S64;
typedef intptr_t iPtrDiff;
#else
typedef unsigned char BYTE;
typedef unsigned short U16;
typedef signed short S16;
typedef unsigned int U32;
typedef signed int S32;
typedef unsigned long long U64;
typedef signed long long S64;
typedef ptrdiff_t iPtrDiff;
#endif
/*-**************************************************************
* Memory I/O
*****************************************************************/
/* MEM_FORCE_MEMORY_ACCESS :
* By default, access to unaligned memory is controlled by `memcpy()`, which is safe and portable.
* Unfortunately, on some target/compiler combinations, the generated assembly is sub-optimal.
* The below switch allow to select different access method for improved performance.
* Method 0 (default) : use `memcpy()`. Safe and portable.
* Method 1 : `__packed` statement. It depends on compiler extension (ie, not portable).
* This method is safe if your compiler supports it, and *generally* as fast or faster than `memcpy`.
* Method 2 : direct access. This method is portable but violate C standard.
* It can generate buggy code on targets depending on alignment.
* In some circumstances, it's the only known way to get the most performance (ie GCC + ARMv6)
* See http://fastcompression.blogspot.fr/2015/08/accessing-unaligned-memory.html for details.
* Prefer these methods in priority order (0 > 1 > 2)
*/
#ifndef MEM_FORCE_MEMORY_ACCESS /* can be defined externally, on command line for example */
# if defined(__GNUC__) && ( defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_6J__) || defined(__ARM_ARCH_6K__) || defined(__ARM_ARCH_6Z__) || defined(__ARM_ARCH_6ZK__) || defined(__ARM_ARCH_6T2__) )
# define MEM_FORCE_MEMORY_ACCESS 2
# elif defined(__INTEL_COMPILER) /*|| defined(_MSC_VER)*/ || \
(defined(__GNUC__) && ( defined(__ARM_ARCH_7__) || defined(__ARM_ARCH_7A__) || defined(__ARM_ARCH_7R__) || defined(__ARM_ARCH_7M__) || defined(__ARM_ARCH_7S__) ))
# define MEM_FORCE_MEMORY_ACCESS 1
# endif
#endif
MEM_STATIC unsigned MEM_32bits(void) { return sizeof(size_t)==4; }
MEM_STATIC unsigned MEM_64bits(void) { return sizeof(size_t)==8; }
MEM_STATIC unsigned MEM_isLittleEndian(void)
{
const union { U32 u; BYTE c[4]; } one = { 1 }; /* don't use static : performance detrimental */
return one.c[0];
}
#if defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==2)
/* violates C standard, by lying on structure alignment.
Only use if no other choice to achieve best performance on target platform */
MEM_STATIC U16 MEM_read16(const void* memPtr) { return *(const U16*) memPtr; }
MEM_STATIC U32 MEM_read32(const void* memPtr) { return *(const U32*) memPtr; }
MEM_STATIC U64 MEM_read64(const void* memPtr) { return *(const U64*) memPtr; }
MEM_STATIC U64 MEM_readST(const void* memPtr) { return *(const size_t*) memPtr; }
MEM_STATIC void MEM_write16(void* memPtr, U16 value) { *(U16*)memPtr = value; }
MEM_STATIC void MEM_write32(void* memPtr, U32 value) { *(U32*)memPtr = value; }
MEM_STATIC void MEM_write64(void* memPtr, U64 value) { *(U64*)memPtr = value; }
#elif defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==1)
/* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */
/* currently only defined for gcc and icc */
#if defined(_MSC_VER) || (defined(__INTEL_COMPILER) && defined(WIN32))
__pragma( pack(push, 1) )
typedef union { U16 u16; U32 u32; U64 u64; size_t st; } unalign;
__pragma( pack(pop) )
#else
typedef union { U16 u16; U32 u32; U64 u64; size_t st; } __attribute__((packed)) unalign;
#endif
MEM_STATIC U16 MEM_read16(const void* ptr) { return ((const unalign*)ptr)->u16; }
MEM_STATIC U32 MEM_read32(const void* ptr) { return ((const unalign*)ptr)->u32; }
MEM_STATIC U64 MEM_read64(const void* ptr) { return ((const unalign*)ptr)->u64; }
MEM_STATIC U64 MEM_readST(const void* ptr) { return ((const unalign*)ptr)->st; }
MEM_STATIC void MEM_write16(void* memPtr, U16 value) { ((unalign*)memPtr)->u16 = value; }
MEM_STATIC void MEM_write32(void* memPtr, U32 value) { ((unalign*)memPtr)->u32 = value; }
MEM_STATIC void MEM_write64(void* memPtr, U64 value) { ((unalign*)memPtr)->u64 = value; }
#else
/* default method, safe and standard.
can sometimes prove slower */
MEM_STATIC U16 MEM_read16(const void* memPtr)
{
U16 val; memcpy(&val, memPtr, sizeof(val)); return val;
}
MEM_STATIC U32 MEM_read32(const void* memPtr)
{
U32 val; memcpy(&val, memPtr, sizeof(val)); return val;
}
MEM_STATIC U64 MEM_read64(const void* memPtr)
{
U64 val; memcpy(&val, memPtr, sizeof(val)); return val;
}
MEM_STATIC size_t MEM_readST(const void* memPtr)
{
size_t val; memcpy(&val, memPtr, sizeof(val)); return val;
}
MEM_STATIC void MEM_write16(void* memPtr, U16 value)
{
memcpy(memPtr, &value, sizeof(value));
}
MEM_STATIC void MEM_write32(void* memPtr, U32 value)
{
memcpy(memPtr, &value, sizeof(value));
}
MEM_STATIC void MEM_write64(void* memPtr, U64 value)
{
memcpy(memPtr, &value, sizeof(value));
}
#endif /* MEM_FORCE_MEMORY_ACCESS */
MEM_STATIC U32 MEM_swap32(U32 in)
{
#if defined(_MSC_VER) /* Visual Studio */
return _byteswap_ulong(in);
#elif defined (__GNUC__)
return __builtin_bswap32(in);
#else
return ((in << 24) & 0xff000000 ) |
((in << 8) & 0x00ff0000 ) |
((in >> 8) & 0x0000ff00 ) |
((in >> 24) & 0x000000ff );
#endif
}
MEM_STATIC U64 MEM_swap64(U64 in)
{
#if defined(_MSC_VER) /* Visual Studio */
return _byteswap_uint64(in);
#elif defined (__GNUC__)
return __builtin_bswap64(in);
#else
return ((in << 56) & 0xff00000000000000ULL) |
((in << 40) & 0x00ff000000000000ULL) |
((in << 24) & 0x0000ff0000000000ULL) |
((in << 8) & 0x000000ff00000000ULL) |
((in >> 8) & 0x00000000ff000000ULL) |
((in >> 24) & 0x0000000000ff0000ULL) |
((in >> 40) & 0x000000000000ff00ULL) |
((in >> 56) & 0x00000000000000ffULL);
#endif
}
MEM_STATIC size_t MEM_swapST(size_t in)
{
if (MEM_32bits())
return (size_t)MEM_swap32((U32)in);
else
return (size_t)MEM_swap64((U64)in);
}
/*=== Little endian r/w ===*/
MEM_STATIC U16 MEM_readLE16(const void* memPtr)
{
if (MEM_isLittleEndian())
return MEM_read16(memPtr);
else {
const BYTE* p = (const BYTE*)memPtr;
return (U16)(p[0] + (p[1]<<8));
}
}
MEM_STATIC void MEM_writeLE16(void* memPtr, U16 val)
{
if (MEM_isLittleEndian()) {
MEM_write16(memPtr, val);
} else {
BYTE* p = (BYTE*)memPtr;
p[0] = (BYTE)val;
p[1] = (BYTE)(val>>8);
}
}
MEM_STATIC U32 MEM_readLE24(const void* memPtr)
{
return MEM_readLE16(memPtr) + (((const BYTE*)memPtr)[2] << 16);
}
MEM_STATIC void MEM_writeLE24(void* memPtr, U32 val)
{
MEM_writeLE16(memPtr, (U16)val);
((BYTE*)memPtr)[2] = (BYTE)(val>>16);
}
MEM_STATIC U32 MEM_readLE32(const void* memPtr)
{
if (MEM_isLittleEndian())
return MEM_read32(memPtr);
else
return MEM_swap32(MEM_read32(memPtr));
}
MEM_STATIC void MEM_writeLE32(void* memPtr, U32 val32)
{
if (MEM_isLittleEndian())
MEM_write32(memPtr, val32);
else
MEM_write32(memPtr, MEM_swap32(val32));
}
MEM_STATIC U64 MEM_readLE64(const void* memPtr)
{
if (MEM_isLittleEndian())
return MEM_read64(memPtr);
else
return MEM_swap64(MEM_read64(memPtr));
}
MEM_STATIC void MEM_writeLE64(void* memPtr, U64 val64)
{
if (MEM_isLittleEndian())
MEM_write64(memPtr, val64);
else
MEM_write64(memPtr, MEM_swap64(val64));
}
MEM_STATIC size_t MEM_readLEST(const void* memPtr)
{
if (MEM_32bits())
return (size_t)MEM_readLE32(memPtr);
else
return (size_t)MEM_readLE64(memPtr);
}
MEM_STATIC void MEM_writeLEST(void* memPtr, size_t val)
{
if (MEM_32bits())
MEM_writeLE32(memPtr, (U32)val);
else
MEM_writeLE64(memPtr, (U64)val);
}
/*=== Big endian r/w ===*/
MEM_STATIC U32 MEM_readBE32(const void* memPtr)
{
if (MEM_isLittleEndian())
return MEM_swap32(MEM_read32(memPtr));
else
return MEM_read32(memPtr);
}
MEM_STATIC void MEM_writeBE32(void* memPtr, U32 val32)
{
if (MEM_isLittleEndian())
MEM_write32(memPtr, MEM_swap32(val32));
else
MEM_write32(memPtr, val32);
}
MEM_STATIC U64 MEM_readBE64(const void* memPtr)
{
if (MEM_isLittleEndian())
return MEM_swap64(MEM_read64(memPtr));
else
return MEM_read64(memPtr);
}
MEM_STATIC void MEM_writeBE64(void* memPtr, U64 val64)
{
if (MEM_isLittleEndian())
MEM_write64(memPtr, MEM_swap64(val64));
else
MEM_write64(memPtr, val64);
}
MEM_STATIC size_t MEM_readBEST(const void* memPtr)
{
if (MEM_32bits())
return (size_t)MEM_readBE32(memPtr);
else
return (size_t)MEM_readBE64(memPtr);
}
MEM_STATIC void MEM_writeBEST(void* memPtr, size_t val)
{
if (MEM_32bits())
MEM_writeBE32(memPtr, (U32)val);
else
MEM_writeBE64(memPtr, (U64)val);
}
/* function safe only for comparisons */
MEM_STATIC U32 MEM_readMINMATCH(const void* memPtr, U32 length)
{
switch (length)
{
default :
case 4 : return MEM_read32(memPtr);
case 3 : if (MEM_isLittleEndian())
return MEM_read32(memPtr)<<8;
else
return MEM_read32(memPtr)>>8;
}
}
#if defined (__cplusplus)
}
#endif
#endif /* MEM_H_MODULE */