[illumos-Developer] Need short help with testing { sed, tr, pax } replacement binaries on i386 ...

[Joerg Schilling]

"I. Szczesniak" <iszczesniak at gmail.com> wrote:

> Joerg, would you kindly check your setup, please? Your numbers
> contradict our findings and real world performance observations on our
> production machines. Our observation is that AST pax is faster than
> star (and a lot more mature, too) and not the opposite as you've
> claimed.
> Something is wrong.

Are you kidding?

All tests that I did run show:

star is always faster than gnutar
star is always faster than /usr/bin/pax
star is always faster than astpax

Las monday I posted numbers for some tests.
As numbers say more than words, let me give add the results for one test:

LD_LIBRARY_PATH=arch/sol11.i386/lib/ arch/sol11.i386/src/cmd/pax/pax -w /usr > /dev/null
12707648 blocks
7:26.145r 14.030u 39.000s 11% 0M 0+0k 0st 0+0io 0pf+0w


gtar -c -Hpax --totals -f- /usr > /dev/zero
gtar: Removing leading `/' from member names
gtar: Removing leading `/' from hard link targets
Total bytes written: 6506311680 (6,1GiB, 17MiB/s)
6:12.198r 6.150u 35.300s 11% 0M 0+0k 0st 0+0io 0pf+0w

Note that gtar wheats if /dev/null is used instead of /devf/zero

/usr/bin/pax -w /usr > /dev/null        
6:01.315r 2.680u 35.060s 10% 0M 0+0k 0st 0+0io 0pf+0w

/usr/bin/pax -w /usr | sdd bs=5k -onull
sdd: Read  1214078 records + 2560 bytes (total of 6216081920 bytes = 6070392.50k).
sdd: Wrote 0 records + 0 bytes (total of 0 bytes = 0.00k).


star -c -Hpax /usr > /dev/null
star: 607027 blocks + 0 bytes (total of 6215956480 bytes = 6070270.00k).
5:44.737r 1.800u 33.830s 10% 0M 0+0k 0st 0+0io 0pf+0w


spax -w /usr > /dev/null
5:45.721r 1.840u 34.290s 10% 0M 0+0k 0st 0+0io 0pf+0w


More meta data:
star -c -Hexustar /usr > /dev/null
star: 635382 blocks + 0 bytes (total of 6506311680 bytes = 6353820.00k).
5:47.788r 3.080u 34.150s 10% 0M 0+0k 0st 0+0io 0pf+0w


Even more meta data:
star -c -Hexustar -dump /usr > /dev/null
star: 635965 blocks + 0 bytes (total of 6512281600 bytes = 6359650.00k).
5:39.062r 3.720u 34.120s 11% 0M 0+0k 0st 0+0io 0pf+0w

Even with archiving much more metadata than others, star is still faster
and still needs less user CPU time that astpax.

Other tests of course show similar results.


And btw. you are of course free to present us your tests for a review.


Jörg

-- 
 EMail:joerg at schily.isdn.cs.tu-berlin.de (home) Jörg Schilling D-13353 Berlin
       js at cs.tu-berlin.de                (uni)  
       joerg.schilling at fokus.fraunhofer.de (work) Blog: http://schily.blogspot.com/
 URL:  http://cdrecord.berlios.de/private/ ftp://ftp.berlios.de/pub/schily