linux <--> windows strcpy etc performance

A

anhadikal

hello,

i have strange results doing some simple "benchmark" ...

on my amd athlon 4800+ x2:

on linux:
-----------------------------------------------------------------------------------------------------------------
cmp_i686
--------------
C strcpy: 3764.7 MB/second (2560.0 MB in
680000 clocks)
our strcpy1: 2976.7 MB/second (2560.0 MB in
860000 clocks)
our strcpy2: 3084.3 MB/second (2560.0 MB in
830000 clocks)
C memcpy: 3084.3 MB/second (2560.0 MB in
830000 clocks)
our memcpy: 1969.2 MB/second (2560.0 MB in
1300000 clocks)

cmp_athlon
-----------------
C strcpy: 3764.7 MB/second (2560.0 MB in
680000 clocks)
our strcpy1: 4000.0 MB/second (2560.0 MB in
640000 clocks)
our strcpy2: 3122.0 MB/second (2560.0 MB in
820000 clocks)
C memcpy: 3122.0 MB/second (2560.0 MB in
820000 clocks)
our memcpy: 1984.5 MB/second (2560.0 MB in
1290000 clocks)
-----------------------------------------------------------------------------------------------------------------

on windows:
-----------------------------------------------------------------------------------------------------------------
C:\DundE\anh\Eigene Dateien\downloads>cmp_athlon.exe
C strcpy: 4821.1 MB/second (2560.0 MB in 531 clocks)
our strcpy1: 4196.7 MB/second (2560.0 MB in 610 clocks)
our strcpy2: 3148.8 MB/second (2560.0 MB in 813 clocks)
C memcpy: 2873.2 MB/second (2560.0 MB in 891 clocks)
our memcpy: 1973.8 MB/second (2560.0 MB in 1297 clocks)

C:\DundE\anh\Eigene Dateien\downloads>cmp_i686.exe
C strcpy: 4812.0 MB/second (2560.0 MB in 532 clocks)
our strcpy1: 2976.7 MB/second (2560.0 MB in 860 clocks)
our strcpy2: 3091.8 MB/second (2560.0 MB in 828 clocks)
C memcpy: 2825.6 MB/second (2560.0 MB in 906 clocks)
our memcpy: 1949.7 MB/second (2560.0 MB in 1313 clocks)
our memcpy: 1949.7 MB/second (2560.0 MB in 1313 clocks)
-----------------------------------------------------------------------------------------------------------------


and now on i5 430m ...

linux:
-----------------------------------------------------------------------------------------------------------------
[dd@lappy Downloads]$ ./cmp_686
C strcpy: 1113.0 MB/second (2560.0 MB in 2300000
clocks)
our strcpy1: 1497.1 MB/second (2560.0 MB in 1710000
clocks)
our strcpy2: 1523.8 MB/second (2560.0 MB in 1680000
clocks)
C memcpy: 1630.6 MB/second (2560.0 MB in 1570000
clocks)
our memcpy: 1207.5 MB/second (2560.0 MB in 2120000
clocks)

[dd@lappy Downloads]$ ./cmp_core2
C strcpy: 1075.6 MB/second (2560.0 MB in 2380000
clocks)
our strcpy1: 1741.5 MB/second (2560.0 MB in 1470000
clocks)
our strcpy2: 1706.7 MB/second (2560.0 MB in 1500000
clocks)
C memcpy: 1600.0 MB/second (2560.0 MB in 1600000
clocks)
our memcpy: 1213.3 MB/second (2560.0 MB in 2110000
clocks)
-----------------------------------------------------------------------------------------------------------------


windows:
-----------------------------------------------------------------------------------------------------------------
C:\Users\Lappi\Desktop\xcp>cmp_686.exe
C strcpy: 3731.8 MB/second (2560.0 MB in 686 clocks)
our strcpy1: 3417.9 MB/second (2560.0 MB in 749 clocks)
our strcpy2: 3417.9 MB/second (2560.0 MB in 749 clocks)
C memcpy: 2562.6 MB/second (2560.0 MB in 999 clocks)
our memcpy: 1823.4 MB/second (2560.0 MB in 1404 clocks)

C:\Users\Lappi\Desktop\xcp>cmp_core2.exe
C strcpy: 4238.4 MB/second (2560.0 MB in 604 clocks)
our strcpy1: 3699.4 MB/second (2560.0 MB in 692 clocks)
our strcpy2: 3459.5 MB/second (2560.0 MB in 740 clocks)
C memcpy: 2552.3 MB/second (2560.0 MB in 1003 clocks)
our memcpy: 2051.3 MB/second (2560.0 MB in 1248 clocks)
-----------------------------------------------------------------------------------------------------------------


I always compiled with gcc 4.5.1 -O2 -s ... Why are the results so
different???


here the bench code:
--------------------------------------------------------------------------------------------------------------------
/****
*
* modified version from Preston L. Bannister
*
**/

#include <stdlib.h>
#include <stdio.h>
#include <time.h>
#include <string.h>


/
*******************************************************************************

*******************************************************************************
* configs

*******************************************************************************

******************************************************************************/
#define LOOPS 10000000
static const char sOut1[] =
"QBTnetfnh8TpTWvPzARBNWr2gMFofe3AzwMXVOGbdL2xOOACwMefrMxpxZ62qakW";
static const char sOut2[] =
"ct6V7lZ42RoryDlvM1EzT54T5qV3DGUA4UIIhVv0TSK0lTx0TKIFc4E4YIdfjfKp";





/
*******************************************************************************

*******************************************************************************
* engine

*******************************************************************************

******************************************************************************/
unsigned int nLength = ::strlen(sOut1);
unsigned int dtLoop = 0;
unsigned int nTotal = 0;
char sWork[256];
typedef void (*doit)(const char*, const char*);

void report_times(const char* s, unsigned int dt)
{
double ts = (double)dt / CLOCKS_PER_SEC;
double mb = (double)(nTotal) / 1000000;
double rate = mb / ts;
printf("%s:\t\t %0.1f MB/second (%0.1f MB in %u clocks)\n", s,
rate, mb, dt);
}

int time_function(doit fn)
{
clock_t t0 = ::clock();
for (int i=0; i<LOOPS; ++i) {
const char* s1 = sOut1 + (15 & i);
const char* s2 = sOut2 + nLength - (15 & i);
(*fn)(s1,s2);
}
return (int):):clock() - t0) - dtLoop;
}

void do_total(const char* s1,const char* s2)
{
nTotal += 4 * nLength;
}
/
*******************************************************************************
* end engine

******************************************************************************/





/
*******************************************************************************

*******************************************************************************
* benchmark

*******************************************************************************

******************************************************************************/
void do_c_strcpy(const char* s1,const char* s2)
{
::strcpy(sWork,s1);
::strcpy(sWork,s2);
}


void our_strcpy1(char* s1,const char* s2)
{
while (*s1++ = *s2++);
}

void our_strcpy2(char* s1,const char* s2)
{
register unsigned int i;

for (i = 0; s2 != 0; ++i)
s1 = s2;
s1 = 0;
}

void do_our_strcpy1(const char* s1,const char* s2)
{
our_strcpy1(sWork,s1);
our_strcpy1(sWork,s2);
}
void do_our_strcpy2(const char* s1,const char* s2)
{
our_strcpy2(sWork,s1);
our_strcpy2(sWork,s2);
}

void do_c_memcpy(const char* s1, const char* s2)
{
int l1 = strlen(s1);
int l2 = strlen(s2);
::memcpy(sWork, s1, l1);
::memcpy(sWork, s2, l2);
}

void our_memcpy(char* dest, const char* src, int size)
{
for(int i=0; i<size; ++i)
dest = src;
}


void do_our_memcpy(const char* s1, const char* s2)
{
int l1 = strlen(s1);
int l2 = strlen(s2);
our_memcpy(sWork, s1, l1);
our_memcpy(sWork, s2, l2);
}

/
*******************************************************************************
* end benchmark

******************************************************************************/





/
*******************************************************************************

*******************************************************************************
* main programm

*******************************************************************************

******************************************************************************/
int main(int ac,char** av)
{
dtLoop = time_function(do_total);

report_times("C strcpy", time_function(do_c_strcpy));
report_times("our strcpy1", time_function(do_our_strcpy1));
report_times("our strcpy2", time_function(do_our_strcpy2));
report_times("C memcpy", time_function(do_c_memcpy));
report_times("our memcpy", time_function(do_our_memcpy));

return 0;
}
 
G

Gene

hello,

i have strange results doing some simple "benchmark" ...

on my amd athlon 4800+ x2:

on linux:
--------------------------------------------------------------------------- --------------------------------------
cmp_i686
--------------
C strcpy:                         3764.7 MB/second (2560.0 MB in
680000 clocks)
our strcpy1:                      2976.7 MB/second (2560.0 MB in
860000 clocks)
our strcpy2:                      3084.3 MB/second (2560.0 MB in
830000 clocks)
C memcpy:                         3084.3 MB/second (2560.0 MB in
830000 clocks)
our memcpy:                       1969.2 MB/second (2560.0 MB in
1300000 clocks)

cmp_athlon
-----------------
C strcpy:                         3764.7 MB/second (2560.0 MB in
680000 clocks)
our strcpy1:                      4000.0 MB/second (2560.0 MB in
640000 clocks)
our strcpy2:                      3122.0 MB/second (2560.0 MB in
820000 clocks)
C memcpy:                         3122.0 MB/second (2560.0 MB in
820000 clocks)
our memcpy:                       1984.5 MB/second (2560.0 MB in
1290000 clocks)
--------------------------------------------------------------------------- --------------------------------------

on windows:
--------------------------------------------------------------------------- --------------------------------------
C:\DundE\anh\Eigene Dateien\downloads>cmp_athlon.exe
C strcpy:                4821.1 MB/second (2560.0 MB in 531 clocks)
our strcpy1:             4196.7 MB/second (2560.0 MB in 610 clocks)
our strcpy2:             3148.8 MB/second (2560.0 MB in 813 clocks)
C memcpy:                2873.2 MB/second (2560.0 MB in 891 clocks)
our memcpy:              1973.8 MB/second (2560.0 MB in 1297 clocks)

C:\DundE\anh\Eigene Dateien\downloads>cmp_i686.exe
C strcpy:                4812.0 MB/second (2560.0 MB in 532 clocks)
our strcpy1:             2976.7 MB/second (2560.0 MB in 860 clocks)
our strcpy2:             3091.8 MB/second (2560.0 MB in 828 clocks)
C memcpy:                2825.6 MB/second (2560.0 MB in 906 clocks)
our memcpy:              1949.7 MB/second (2560.0 MB in 1313 clocks)
our memcpy:              1949.7 MB/second (2560.0 MB in 1313 clocks)
--------------------------------------------------------------------------- --------------------------------------

and now on i5 430m ...

linux:
--------------------------------------------------------------------------- --------------------------------------
[dd@lappy Downloads]$ ./cmp_686
C strcpy:                1113.0 MB/second (2560.0 MB in 2300000
clocks)
our strcpy1:             1497.1 MB/second (2560.0 MB in 1710000
clocks)
our strcpy2:             1523.8 MB/second (2560.0 MB in 1680000
clocks)
C memcpy:                1630.6 MB/second (2560.0 MB in 1570000
clocks)
our memcpy:              1207.5 MB/second (2560.0 MB in 2120000
clocks)

[dd@lappy Downloads]$ ./cmp_core2
C strcpy:                1075.6 MB/second (2560.0 MB in 2380000
clocks)
our strcpy1:             1741.5 MB/second (2560.0 MB in 1470000
clocks)
our strcpy2:             1706.7 MB/second (2560.0 MB in 1500000
clocks)
C memcpy:                1600.0 MB/second (2560.0 MB in 1600000
clocks)
our memcpy:              1213.3 MB/second (2560.0 MB in 2110000
clocks)
--------------------------------------------------------------------------- --------------------------------------

windows:
--------------------------------------------------------------------------- --------------------------------------
C:\Users\Lappi\Desktop\xcp>cmp_686.exe
C strcpy:                3731.8 MB/second (2560.0 MB in 686 clocks)
our strcpy1:             3417.9 MB/second (2560.0 MB in 749 clocks)
our strcpy2:             3417.9 MB/second (2560.0 MB in 749 clocks)
C memcpy:                2562.6 MB/second (2560.0 MB in 999 clocks)
our memcpy:              1823.4 MB/second (2560.0 MB in 1404 clocks)

C:\Users\Lappi\Desktop\xcp>cmp_core2.exe
C strcpy:                4238.4 MB/second (2560.0 MB in 604 clocks)
our strcpy1:             3699.4 MB/second (2560.0 MB in 692 clocks)
our strcpy2:             3459.5 MB/second (2560.0 MB in 740 clocks)
C memcpy:                2552.3 MB/second (2560.0 MB in 1003 clocks)
our memcpy:              2051.3 MB/second (2560.0 MB in 1248 clocks)
--------------------------------------------------------------------------- --------------------------------------

I always compiled with gcc 4.5.1 -O2 -s ... Why are the results so
different???

here the bench code:
--------------------------------------------------------------------------- -----------------------------------------
/****
 *
 * modified version from Preston L. Bannister
 *
 **/

#include <stdlib.h>
#include <stdio.h>
#include <time.h>
#include <string.h>

/
*************************************************************************** ****

*************************************************************************** ****
 * configs

*************************************************************************** ****

*************************************************************************** ***/
#define LOOPS 10000000
static const char sOut1[] =
"QBTnetfnh8TpTWvPzARBNWr2gMFofe3AzwMXVOGbdL2xOOACwMefrMxpxZ62qakW";
static const char sOut2[] =
"ct6V7lZ42RoryDlvM1EzT54T5qV3DGUA4UIIhVv0TSK0lTx0TKIFc4E4YIdfjfKp";

/
*************************************************************************** ****

*************************************************************************** ****
 * engine

*************************************************************************** ****

*************************************************************************** ***/
unsigned int nLength = ::strlen(sOut1);
unsigned int dtLoop = 0;
unsigned int nTotal = 0;
char sWork[256];
typedef void (*doit)(const char*, const char*);

void report_times(const char* s, unsigned int dt)
{
    double ts = (double)dt / CLOCKS_PER_SEC;
    double mb = (double)(nTotal) / 1000000;
    double rate = mb / ts;
    printf("%s:\t\t %0.1f MB/second (%0.1f MB in %u clocks)\n", s,
rate, mb, dt);

}

int time_function(doit fn)
{
    clock_t t0 = ::clock();
    for (int i=0; i<LOOPS; ++i) {
        const char* s1 = sOut1 + (15 & i);
        const char* s2 = sOut2 + nLength - (15 & i);
        (*fn)(s1,s2);
    }
    return (int):):clock() - t0) - dtLoop;

}

void do_total(const char* s1,const char* s2)
{
    nTotal += 4 * nLength;}

/
*************************************************************************** ****
 * end engine

*************************************************************************** ***/

/
*************************************************************************** ****

*************************************************************************** ****
 * benchmark

*************************************************************************** ****

*************************************************************************** ***/
void do_c_strcpy(const char* s1,const char* s2)
{
    ::strcpy(sWork,s1);
    ::strcpy(sWork,s2);

}

void our_strcpy1(char* s1,const char* s2)
{
    while (*s1++ = *s2++);

}

void our_strcpy2(char* s1,const char* s2)
{
    register unsigned int i;

    for (i = 0; s2 != 0; ++i)
        s1 = s2;
    s1 = 0;

}

void do_our_strcpy1(const char* s1,const char* s2)
{
    our_strcpy1(sWork,s1);
    our_strcpy1(sWork,s2);}

void do_our_strcpy2(const char* s1,const char* s2)
{
    our_strcpy2(sWork,s1);
    our_strcpy2(sWork,s2);

}

void do_c_memcpy(const char* s1, const char* s2)
{
    int l1 = strlen(s1);
    int l2 = strlen(s2);
    ::memcpy(sWork, s1, l1);
    ::memcpy(sWork, s2, l2);

}

void our_memcpy(char* dest, const char* src, int size)
{
    for(int i=0; i<size; ++i)
        dest = src;

}

void do_our_memcpy(const char* s1, const char* s2)
{
    int l1 = strlen(s1);
    int l2 = strlen(s2);
    our_memcpy(sWork, s1, l1);
    our_memcpy(sWork, s2, l2);

}

/
*************************************************************************** ****
 * end benchmark

*************************************************************************** ***/

/
*************************************************************************** ****

*************************************************************************** ****
 * main programm

*************************************************************************** ****

*************************************************************************** ***/
int main(int ac,char** av)
{
    dtLoop = time_function(do_total);

    report_times("C strcpy", time_function(do_c_strcpy));
    report_times("our strcpy1", time_function(do_our_strcpy1));
    report_times("our strcpy2", time_function(do_our_strcpy2));
    report_times("C memcpy", time_function(do_c_memcpy));
    report_times("our memcpy", time_function(do_our_memcpy));

    return 0;
}
Click to expand...


Well, for one thing, you measuring actual elapsed time while the OS is
consuming time as well. Difference OSs consume different amounts.
You didn't say if you're running both OSs on the same machine and
how. If you're using a hypervisor, there's overhead. If on different
hardware, the memory systems can have a big effect on raw copy
performance.
 
E

Eric Sosman

hello,

i have strange results doing some simple "benchmark" ...
[... results snipped ...]

I always compiled with gcc 4.5.1 -O2 -s ... Why are the results so
different???
Click to expand...

One thing that stands out is the huge difference in the number
of "clocks" the two systems report: The Linux counts are six and seven
digits long (with many trailing zeroes), while Windows reports times
of three or four digits (with variation in the units' place). The two
are clearly not measuring time the same way, and they may not even be
measuring the same thing.
here the bench code:
[... code snipped ...]
Click to expand...

It doesn't appear to be C; I'm guessing it's a language with some
degree of shared ancestry, a gorilla to C's chimpanzee. If it were C,
I'd draw your attention to

1) The `nTotal' variable is not the amount of data read and written,
by the functions under test, but somewhat larger. The "MB/second"
number that derives from it is therefore suspect.

2) It does not appear that any attempt has been made to account
for the time consumed by the test framework itself. The clock count
includes not only the data-copying time, but also the loop overhead,
the function linkages, and so on. Again, "MB/second" is suspect.

3) It does not appear that any attempt has been made to equalize
cache effects between the various tests. It is quite possible that
the first function tested will pay the penalty for getting data from
RAM to cache, while subsequent functions find the cache already hot.

4) This may be what you intended, but notice that memcpy() and
its variants are asked to do less work than strcpy() and its.

5) There's a lot of sloppiness with types, the author apparently
believing that `int' is good enough for everything. The clock_t and
size_t types are there for a reason, and the program would be more
trustworthy if it used them. As things stand, a system with (for
instance) 32-bit `int' and 64-bit `clock_t' may give hopelessly
bogus results.

But, as noted, the code is not C so all these observations may
be for nothing. Perhaps a visit to comp.lang.c++ might elicit some
better-informed opinions.
 
I

Ian Collins

Well, for one thing, you measuring actual elapsed time while the OS is
consuming time as well. Difference OSs consume different amounts.
You didn't say if you're running both OSs on the same machine and
how. If you're using a hypervisor, there's overhead. If on different
hardware, the memory systems can have a big effect on raw copy
performance.
Click to expand...

Did you have to quote the entire post just to add that?

The code isn't C and the methods used suspect, what more is there to add?
 
E

Eric Sosman

Eric Sosman said:
[...]
It doesn't appear to be C; I'm guessing it's a language with some
degree of shared ancestry, a gorilla to C's chimpanzee.
Click to expand...
[...]

It's nearly C, though. The only changes I had to make to get it
to compile without warning (using "gcc -std=c99") were:

Re-join some comment lines that had been split.

Delete all "::" tokens.

Change
unsigned int nLength = strlen(sOut1); -
to
unsigned int nLength = sizeof sOut1 + 1;
Click to expand...

ITYM "- 1". But the point remains: If he's using Objective COBOL,
he should ask his questions in an Objective COBOL forum and not here.
Plus, the other points I raised would remain (or "become") relevant if
his Subjective FORTRAN were translated to C.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Similar Threads

Decompressed bitmap image doesn't properly render when using WinGDI 2
Adding adressing of IPv6 to program 1
[XPOST] [C] Access speed: pointers Vs array 4
strccpy 7
Anything wrong with this function? 9
Do you like my strlcpy strlcat etc? 13
reading a structure with pointer 4
String Manipulation Functions - strcpy, strncpy 4

Members online

No members online now.
Total: 287 (members: 1, guests: 286)
Robots: 286

Forum statistics

Threads
473,954
Messages
2,570,116
Members
46,704
Latest member
BernadineF

Latest Threads

Top