opensolaris C library vs. GNU C Library

S

smnoff

I have taken a look at the source code for a few string functions like
memcpy() and strstr() in OpenSolaris and GNU and I find them somewhat
different.

The GNU version of strstr() seems like a mess and very cluttered. Not sure
how fast it is but the OpenSolaris version of strstr() seems
straightforward.

Any reason why this is?

Also any comments, recommendation, stuff a newbie should know between the
two libraries,
OpenSolaris and GNU?


Open Solaris
http://cvs.opensolaris.org/source/xref/on/usr/src/common/util/string.c

Free Software Foundation (GNU)
http://wdiff.progiciels-bpi.ca/showfile.html?name=dist/lib/strstr.c
 
T

tedu

smnoff said:
The GNU version of strstr() seems like a mess and very cluttered. Not sure
how fast it is but the OpenSolaris version of strstr() seems
straightforward.

Any reason why this is?

* I deliberately chose not to comment it. You should have at least
* as much fun trying to understand it, as I had to write it :).

i think that pretty much says it all.
Also any comments, recommendation, stuff a newbie should know between the
two libraries,
OpenSolaris and GNU?

i'd recommend against adopting a similar philosophy when writing your
own code.
 
J

jacob navia

tedu a écrit :
* I deliberately chose not to comment it. You should have at least
* as much fun trying to understand it, as I had to write it :).

i think that pretty much says it all.


Yes. You are right. Most of the GNU code is like that.

Messy, and without a single comment, beides the license
stuff.

Open source means nothing at all since to understand
the stuff you have to spend more time than if you start
from scratch.

I wrote this:

extern int _stdcall GetTickCount(void);

int main(void)
{
char buf[65535];
char *p="Search";
int t,i;

memset(buf,'A',sizeof(buf)-1);
buf[sizeof(buf)-1] = 0;
strcpy(buf+sizeof(buf)-sizeof("Search")-1,p);
t = GetTickCount() ;
for (i=0; i<10000;i++) {
GNUstrstr(buf,"Search");
}
t = GetTickCount() - t;
printf("GNU=%d ms\n",t);
t = GetTickCount() ;
for (i=0; i<10000;i++) {
Sunstrstr(buf,"Seach");
}
t = GetTickCount() - t;
printf("Sun = %d ms\n",t);
}

The output is:
GNU=1313 ms
Sun = 1640 ms

If we change the strcpy line to more realistic test
conditions:
memcpy(buf+1000,p,6);

results change to:

GNU=16 ms
Sun = 31 ms

This means that both algorithms converge with large strings, what is not
surprising since input/output from/to main memory becomes the dominant
factor of the computation, and both algorithms are then the same.

With shorter strings GNU is twice as fast, but since the strings are
shorter the absolute difference is less (only 15ms after 10 thousand
runs!) so it is probably not worth the effort of understanding
the GNU version, full of goto's and a very complex control flow.

Surely I would not love being the maintainer of that stuff.

The size of the GNU function is 260 bytes, the Sun version is
only 110 bytes.

The GNU version would be important when strstr makes more than 90%
of the run time of your program. For all other situations Sun's version
is the same

jacob
 
J

jacob navia

jacob navia a écrit :
I wrote this:

extern int _stdcall GetTickCount(void);

int main(void)
{
char buf[65535];
char *p="Search";
int t,i;

memset(buf,'A',sizeof(buf)-1);
buf[sizeof(buf)-1] = 0;
// Prepare a string of 64K having the searched
// for string at the end of the string, i.e.
// almost at the end of the 64K
strcpy(buf+sizeof(buf)-sizeof("Search")-1,p);
t = GetTickCount() ;
for (i=0; i<10000;i++) {
GNUstrstr(buf,"Search");
}
t = GetTickCount() - t;
printf("GNU=%d ms\n",t);
t = GetTickCount() ;
for (i=0; i<10000;i++) {
Sunstrstr(buf,"Seach");
}
t = GetTickCount() - t;
printf("Sun = %d ms\n",t);
}


Excuse me. I was complaining about lack of comments in GNU stuff
and wrote a piece of code without any comment !!!!!

I hope is clearer now.

jacob
 
W

William Ahern

The GNU version would be important when strstr makes more than 90% of the
run time of your program. For all other situations Sun's version is the
same

Except, with a poor library every time you re-implement and optimize one
function, another one will rise to the top. I'd rather a well honed
library where I didn't have to worry about such things and could focus on
my own code. So, if this difference were isolated you have a point, if not
shame on Sun.

Practically speaking, you want simple code and simple algorithms on the
edges of your software, where you spend the vast majority of your time
changing behavior.
 
D

Dave Thompson

On Fri, 09 Jun 2006 00:16:34 +0200, jacob navia
char buf[65535];
char *p="Search";
int t,i;

memset(buf,'A',sizeof(buf)-1);
buf[sizeof(buf)-1] = 0;
strcpy(buf+sizeof(buf)-sizeof("Search")-1,p);

If you made it char p [] = "Search" could use sizeof (p).
t = GetTickCount() ;
for (i=0; i<10000;i++) {
GNUstrstr(buf,"Search");
}
t = GetTickCount() - t;
printf("GNU=%d ms\n",t);
t = GetTickCount() ;
for (i=0; i<10000;i++) {
Sunstrstr(buf,"Seach");
}
t = GetTickCount() - t;
printf("Sun = %d ms\n",t);

I hope you actually searched for Search not Seach in the Solaris case
to be (completely) fair. Or to avoid such typos, use p in both cases.


- David.Thompson1 at worldnet.att.net
 
W

websnarf

jacob said:
This means that both algorithms converge with large strings, what is not
surprising since input/output from/to main memory becomes the dominant
factor of the computation, and both algorithms are then the same.

This will probably not be true until you start swapping to disk.
Memory is still O(n), but is mitigated by bandwidth capabilities of
your hardware while the O(n) algorithm of the strstr() itself is still
dependent on the efficiency of the algorithm.
With shorter strings GNU is twice as fast, but since the strings are
shorter the absolute difference is less (only 15ms after 10 thousand
runs!) so it is probably not worth the effort of understanding
the GNU version, full of goto's and a very complex control flow.

Because you're going to get so much more understanding out of "you
can't shift a negative number on some platforms". It *IS* worth a
minute or two of study, because the author didn't comment it, yet it
achieves better performance. If you figure it out, you may be able to
achieve that performance without the non-maintainability penality. You
instead prefer to stay in the dark for lack of wanting to look at a 100
lines of code?
Surely I would not love being the maintainer of that stuff.

In the case of strstr(), or other well defined library functions, its
actually quite easy to maintain it, regardless of how convoluted it is.
Just write tests for it.
The size of the GNU function is 260 bytes, the Sun version is
only 110 bytes.

The GNU version would be important when strstr makes more than 90%
of the run time of your program. For all other situations Sun's version
is the same

Yes, but *understanding* the code (for its performance) is the most
important alternative of all. Doing so allows you do this:

char * QEDstrstr (const char * d0, const char * d1) {
int i, j;
char c0, c1;

/* peel off case: strstr (*, "") */
if ('\0' == (c0 = d1[0])) return (char *) d0;

/* peel off case: strstr (*, one-character-string) */
if ('\0' == (c1 = d1[1])) {
char * d = (char *) d0;
for (; c0 != *d; d++) {
if ('\0' == *d) return NULL;
}
return d;
}

for (i = 0;; i++) {

/* Unrolled (once) first character test */
loop0:;
if (c0 != d0) {
if ('\0' == d0) return NULL;
if (c0 != d0[i+1]) {
if ('\0' == d0[i+1]) return NULL;
i+=2;
goto loop0;
}
i++;
}

/* Second character test */
if (c1 != d0[i+1]) continue;

/* Check beyond first two characters */
for (j = 2; d1[j] == d0[i+j]; j++) {
if ('\0' == d1[j]) return (char *) (d0 + i);
}
if ('\0' == d1[j]) return (char *) (d0 + i);
}
}

This achieves very close to the GNUstrstr performance (from my own
testing, on multiple compilers), while being somewhat more readable.
The trick, of course, was to unroll (the core reason why the GNU
version is faster), and peel off one case (which is just a standard
thing I do, as necessary). Its commented, and therefore even more
maintainable than the Sun version.

Now, would you say, this sort of optimization was not worth it, because
I cannot possibly optimize for multiple platforms/compilers at once
(even though I did just that)? That the micro-technique (unrolling) is
not worth it because you're hoping that your compiler will do it for
you (only one I tried succeeded in doing that)?
 

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

Members online

Forum statistics

Threads
473,769
Messages
2,569,582
Members
45,057
Latest member
KetoBeezACVGummies

Latest Threads

Top