Skip to main content.
home | support | download

Back to List Archive

Re: benchmarking compression versus no compression

From: Bill Moseley <moseley(at)not-real.hank.org>
Date: Thu Sep 11 2003 - 00:55:15 GMT
On Wed, Sep 10, 2003 at 03:42:37PM -0700, Aaron Bazar wrote:
> Hello everybody!
> 
> 
> Has anybody done any tests or benchmarks on using swish-e with zlib and
> without? Does compression affect the speed at all or not very much?

On your Apple II or your SCSI RAID 5 array?  P133 or P4 2.4?

You have to run your own tests.


$ ls -l index.swish-e.prop  nocompress.prop
-rw-r--r--    1 moseley  moseley   3018375 2003-09-10 12:38 index.swish-e.prop
-rw-r--r--    1 moseley  moseley   7741872 2003-09-10 17:41 nocompress.prop

1809 documents.


$ time swish-e -w not dkdkd -p swishdescription > /dev/null

real    0m3.940s
user    0m3.850s
sys     0m0.080s

$ time swish-e -w not dkdkd -p swishdescription -f nocompress > /dev/null

real    0m3.645s
user    0m3.560s
sys     0m0.070s

Keep in mind that's dumping a lot of output:

$ swish-e -w not dkdkd -p swishdescription -f nocompress |wc -c
7740812.

But normally don't print all 1809 results/7MB data:  Try a page of 20:

$ time swish-e -w not dkdkd -p swishdescription -m20 > /dev/null

real    0m0.147s
user    0m0.130s
sys     0m0.000s

$ time swish-e -w not dkdkd -p swishdescription -f nocompress -m20 > /dev/null 

real    0m0.141s
user    0m0.130s
sys     0m0.010s

Try sorting:

$ time swish-e -w not dkdkd -s swishdescription  -m20 >/dev/null

real    0m0.083s
user    0m0.070s
sys     0m0.010s

$ time swish-e -w not dkdkd -f nocompress -s swishdescription  -m20 >/dev/null 

real    0m0.084s
user    0m0.080s
sys     0m0.000s


-- 
Bill Moseley
moseley@hank.org
Received on Thu Sep 11 00:55:35 2003