View Full Version : Tweaking SlimServer weak WLANs, slow CPUs, OGG library

Andrew Lucas
2004-01-15, 06:20

This is really interesting.

I was considering a Via Epia 10000 (low power consumption is attractive) to
serve two squeezebox's and one Slimp3 using FLAC with XP but I am worried
that it may not have enough horsepower, I guess your configuration is very
different but any thoughts on this? Or if anyone else has tried this I would
be interested to hear how well it worked.


>From: John Ellis <john.ellis (AT) ise-indy (DOT) com>
>Reply-To: Slim Devices Discussion <discuss (AT) lists (DOT) slimdevices.com>
>To: discuss (AT) lists (DOT) slimdevices.com
>Subject: [slim] Tweaking SlimServer weak WLANs, slow CPUs, OGG library
>Date: Wed, 14 Jan 2004 21:54:09 -0500
>Hash: SHA1
>Just thought I'd offer advice on how to setup a Squeezebox with really,
>really minimum requirements. Here's my original setup:
>o Ogg Vorbis files - 288 albums and 3812 tracks
>o VIA Eden 5000 MiniITX server, 40 GB HD, 64 MB RAM
>o 802.11b WLAN (WAP at the other side of the house)
>Tracks skipped like mad this way. CPU usage was pegged at 100%, my
>wireless signal was down to about 40% (less with the microwave heatin' a
>burrito). Bear in mind a VIA Eden 5000 is pretty much a Cyrix x586.
>Remember those things?
>So I did the following things:
>o Upgraded by 256 MB RAM to 320 MB (helped a lil, mainly with the number
>of processes running at the same time)
>o Got a directional antenna for the Squeezebox. This helped bring my
>wireless signal up to 70-80% (big difference). The one I picked up was a
>Trendware 6dBi Directional Antenna.
>o Got the latest version of vorbis tools (1.0.1) and compiled them
>specifically for my x586 platform. Helped a little big - oggdec only
>takes 10-12% of my processor time.
>o Got the latest version of lame (3.95) and uber-compiled it for my
>platform. I used the optimizations in lame.spec (included in the package
>for RPM building) and did some major GCC optimization. Made a bit of a
>o Added the flags '-m s' and '-f' to lame. The big difference was with
>'-f' - it takes away some extra acoustic algorithms used for encoding
>that I really don't miss all that much. Using '-m s' forced stereo mode
>instead of joint stereo mode, which gave me some additional speed. This
>brought lame down from 95% processor utilization to 50% utilization. Big
>Hope it might help others...
>- --
>John T. Ellis
>Information Systems Experts, LLC
>Version: GnuPG v1.2.2-rc1-SuSE (GNU/Linux)
>iD8DBQFABgDTRbpfi9ylWxoRAjR1AJ993aHWR6jrPKdFOWzaD2 E/Mi67HgCcCxAP