Very Computer

The problem with D2D is the disk I/O. Recording CD quality sound requires
that the file system+disk I/O code is able to store continuously 176k of
data to a file without pauses longer than 0.3 sec. On my system this is
not possible (DX4/100+IDE disks+ext2fs). It may work with SCSI disks
and/or using raw disk partitions (no fs on them). Doing recording and
playback at the same time makes things still worse.

Quote:>I looked the sound driver sources, and noticed that it is only designed
>to do one of these jobs at a time. I think that the major reason for this
>is that most of the cards have only 1 DMA channel.

Quote:>> Do you think that an interrupt driven driver is a possible solution?
>> Would it be slow? (for a 44khz stereo and bidirectionnal transfert,
>would it take less than 5% cpu time on a pentium 60)

Quote:>I currently own a Sound Galaxy nx pro 16 sound card, which is a
>8 bits sound blaster compatible plus a Windows Sound System 16 bits
>device (ad1848). In the driver source (ad1848.c), I noticed that there
>is a mode with two DMA channels. Is it hardware compatible? Can it work?

Hannu
--
-----------------------------
Hannu Savolainen

"Hackers don't use factory prebooted DOS."

: The problem with D2D is the disk I/O. Recording CD quality sound requires
: that the file system+disk I/O code is able to store continuously 176k of
: data to a file without pauses longer than 0.3 sec. On my system this is
: not possible (DX4/100+IDE disks+ext2fs). It may work with SCSI disks
: and/or using raw disk partitions (no fs on them). Doing recording and
: playback at the same time makes things still worse.

No problem on my system, (Cyrix 486dx2/80, PCI ncr53c810 SCSI, Seagate
ST12400N Hawk II 2 gig disk, SB16 for in and GUS oldie for out,) I can even
play one 44.1k/16bit file while recording another, as long as the system
is doing _NOTHING ELSE_.

The hard part is trying to do useful work while also going back to the read
every 0.18 seconds. Today (Sunday) I'm going to try to implement a 128k
secondary buffer in dmabuf.c, which should allow 0.74 seconds between reads.
I implemented select last night, maybe the two together will allow me to get
something done.

: >I looked the sound driver sources, and noticed that it is only designed
: >to do one of these jobs at a time. I think that the major reason for this
: >is that most of the cards have only 1 DMA channel.
: This limitation is there since at the time I started writing the sound
: driver there was just the SB which was not able to work in duplex mode.
: However since full duplex is the most requested feature, it's propably
: the time to start implementing it.

Yay!

JJH
--
 I like to believe that people in the long run are going to do more to
promote peace than our governments.  Indeed, I think that people want
peace so much that one of these days governments had better get out of
the way and let them have it.
                -- Dwight D. Eisenhower