[rescue] Sparc Classic

[Greg A. Woods]

[ On Tuesday, January 15, 2002 at 14:40:51 (-0500), Kurt Mosiejczuk wrote: ]
> Subject: Re: [rescue] Sparc Classic
>
> What OS and how did you compile mpg123?

NetBSD of course [:-)], with the in-tree gcc, which is apparently
egcs-1.1.2 in the release I'm using.

> Linux has horrible latency for a lot of things.  And sparc audio
> requires a lot of handholding by the OS.  Skipping might make it
> seem like the machine can't keep up FP-wise, but it's more likely
> to be the OS can't keep the sound chip fed.

Like I said, I'm not using sparc audio.  Only sparc CPU.  And a not
otherwise too shabby a one at that:

cpu0 at mainbus0: TMS390Z50 v0 or TMS390Z55 @ 75 MHz, on-chip FPU
cpu0: physical 20K instruction (64 b/l), 16K data (32 b/l), 1024K external (32 b/l): cache enabled

I can't even begin to use sparc audio on this machine as there's no
driver for it yet:

SUNW,DBRIe at sbus0 slot 15 offset 0x8010000 level 9 not configured


The OS can certainly keep the CPU fed with data from the disk, and
there's no performance problem with sending the raw audio out on the
network to the NCD.  (not even when reading it from the not-so-fast SCSI
drive I have, and certainly not when simultaneously reading a measly
128-kbit mp3 stream from the network via TCP)  An SS-20 is a very
powerful computer, especially in terms of I/O throughput (given the
hardware interfaces it has to work with, of course) -- it can squeeze
more performance out of the same peripherals than most any other system
of its generation.  I've got 320MB of RAM in this machine too, so
there's no danger it was wasting any time paging or anything silly like
that.

Once I finally get NAS to compile again (this time I'm building in an
all-static environment and I don't think anyone's ever done that with
NAS-1.4.2!) I'll rebuild mpg123 with the built-in NAS support and see if
avoiding the "|sox|auplay" processes improves things (though unless sox
was doing a lot of the work, I kinda doubt it).

I'll have to build ogg123 to compare I guess.  As I recall it uses a lot
less CPU.

-- 
								Greg A. Woods

+1 416 218-0098;  <gwoods at acm.org>;  <g.a.woods at ieee.org>;  <woods at robohack.ca>
Planix, Inc. <woods at planix.com>; VE3TCP; Secrets of the Weird <woods at weird.com>