Lets see if the RightMark results deteriorate with increased CPU usage
(all done with WiFi @ 60% strength, EDO kernel, analogue output
measured):
14431

Far left, Touch playing at 24/96 SERVER SIDE DECODING, default "Now
Playing".

2nd left, Touch playing 24/96 SERVER SIDE DECODING, with Spectrum
Analyzer "Now Playing".

3rd left, Touch playing 24/96 SERVER SIDE DECODING, Spectrum Analyzer +
5 TTY sessions running "nice -n -20 top -d1".
- This means that there are 5 PuTTY sessions open, each one running
'top' at a high priority level and refreshing system status every
second. With server side decoding of the FLAC, this is the limit for me
at WiFi 60%. If I open another PuTTY, this results in constant
rebuffering errors (all the PuTTY's needing to send out a screen of data
every second).

4 & 5th left, Touch playing 24/96 NATIVE FLAC, Spectrum Analyzer + 5 or
10 TTY sessions with "nice -n -20 top -d1".
- With native FLAC decode, the WiFi has more room to "breathe" due to
data compression and can handle more TTY sessions without buffering
errors!

As you can see, the measured results are rock solid even with an
unreasonable number of TTY sessions open (idle CPU usage went from ~1-5%
to 30% with all these PuTTY sessions). What this tells me is the Linux
realtime scheduler in the Touch is robust. For example, lower priority
processes like the Spectrum Analyzer refreshes slower as the number of
TTY/top sessions increases and the GUI responsiveness goes down as well
but the sound quality remains unchanged as far as I can tell with
measurements.

Here's what my screen looked like with all these TTY sessions all logged
into the Touch:
14432

As for 24/192, via ethernet I just listened to Cat Steven's "Tea For The
Tillerman" (HDTracks), and now starting Neil Young's "Harvest" (DVD-A
rip) with 10 TTY sessions in the background running the high-priority
'top' with Spectrum Analyzer display. No audible problems with the Touch
connected to the Essence One by TosLink. I didn't bother measuring this
since it sounds great and I didn't feel like dragging out the XLR cables
again ;-)!

Bottom line: I still see no evidence of high CPU use causing significant
change to the audio output...  If there is something, it must be below
the noise floor of the Touch DAC in any case. Likewise, the Touch seems
to have good headroom for 24/192 playback as well. IMO, it's more
important then to make sure there's enough data bandwidth (ie. use
ethernet when you can!) than worry about CPU usage causing audio
degradation.

Usual disclaimer of course is that the measurements are with test
signals which may not represent the same performance with real music.


+-------------------------------------------------------------------+
|Filename: Multiuser_Touch!.jpg                                     |
|Download: http://forums.slimdevices.com/attachment.php?attachmentid=14432|
+-------------------------------------------------------------------+

------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=97950

_______________________________________________
audiophiles mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/audiophiles

Reply via email to