Re: gbde Performance - 35Mb/s vs 5.2 MB/s

From: Heiko Schaefer <hschaefer_at_fto.de>
Date: Mon, 26 May 2003 14:25:08 +0200 (CEST)
Hi Christophe,

> I  get 35MB/s write without gbde and 5 MB on a gbde crypted device.
>
> Hardware:
> 700 GB ATA Raid5 3ware, P4 2.6ghz , 512mb ram
>
> I set sector size to 2048, performance increased to 5mb/sec. cmd to
> creat fs was:
>
> newfs -b 16384 -f 2048 -U -m 0.0005% -i 262144 /dev/twed0s1h.bde

Poul gave me the following tip on this list in a mail on Tue, 29 Apr 2003:

"Remember to set the sectorsize in gbde (gbde init -i) to the fragment
size of your filesystem (typically 2048 for ufs), this is critical
for performance."

did you do that?

i do not have exact numbers, but i think i can get approx 10MB/s writing
on my gbde disks with an athlon xp 1800+.

...i would be curious about what top (or ps auxw) tell you while you use
your gbde disk. from what i've experienced, your cpu shouldn't be maxed
out with 5MB/s throughput.

> Could anyone tell me if this is supposed to be slow like that ? Or is
> there anything I missed ?

gbde is fairly slow for me too, but by using "gbde init -i", the cpu is
the limiting factor. and i'm fairly fine with that, assuming the crypto
that is done is sound :) ...unfortunately only with time that will become
somewhat certain.

i am still (neverending story :( ) struggling with hardware, but one of
these days i plan to post some numbers :) it seems, however that my cpu is
theoretically capable of processing data with gbde at a rate of approx
25MB/s (that is an extrapolated number from "ps auwx" output for the gbde
process and the actual throughput).

cheers,

Heiko

-- 
Free Software. Why put up with inferior code and antisocial corporations?
http://www.gnu.org/philosophy/why-free.html
Received on Mon May 26 2003 - 03:25:14 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:09 UTC