for pd-0.26 - 26 Channels I/O

Winfried Ritsch ritsch at iem.mhsg.ac.at
Mon Aug 23 23:17:49 CEST 1999


hello,

thanx for pd-0.26..

because I have been asked ... I just have finished the first version
of the linux driver for the RME DIGI9652 (Hammerfall) Card which is a
PCI Card that offers 3 ADAT-IOs and one SPDIF/AESEBU + Word Clock and
ADAT TimeCode (see http://www.rme-audio.de/) I have it in use with
Fostex VC-8s and it works great.

The driver is a bit OSS-compatible but mainly a native version
especcially for the card. (since multichannel and 24Bit are not
directly supported by OSS)

I made a first version of integration in pd-0.26, which works fine,
but does not have the best performance. This will change in near
future doing a memory mapped version, since the card does master-mode pci
and only with memory mapped interface you get the best performance
(No CPU-Power if no copying). 

I ask for people to test this with other configurations, since I think
with my computer (PII-400MHz) and the digi9652 and a RIVA-TNT 32MB RAM
Video-Card the bus is very much stressed and RAM access becomes slower
;-(, but I cant tell excatly ... (miller, how did you do the fast
PCI128 driver thing ;-) I cant get the same latency, ...).
 
the test versions are unter http://iem.mhsg.ac.at/programmier/

I will be used in a 24 Channel Installation at OK Linz during
ARS-Electronica festival in Linz in two weeks. After that I will
release a the mmapped version (hopefully working).

It is planned to make an ALSA Driver out of this, but native support 
will stay.

mfg winfried



More information about the Pd-list mailing list