Re[4]: [linux-audio-dev] Reverse-engineering files

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: Re[4]: [linux-audio-dev] Reverse-engineering files
From: Rick Burnett (destinytech_AT_spacey.net)
Date: Fri Dec 01 2000 - 09:23:03 EET


Hehe, I agree :) I wasn't implying a complete software solution for
Emu, I was just meaning to say I think they are afraid to push more of
it to the PC side because it is then easier for other companies to
develop the same thing :)

I use a SBLive. The problem with the sound cards is Creative is
notorious for not using all the functionality of the chips. For
instance, on the Waveblaster you couldn't use the built in effects of
the proteus properly, they physically hard wired some of the pins to
disable certain features. Not sure if it was for time reasons, cost,
or just to make sure the Proteus dedicated units would have some
difference :)

I agree on the part about chip development. What I see alot of is
companies creating a new hybrid chip to do things that they want. It
is cheaper for a company that plans on making a high percentage of
units to develop their own chip if they have special needs and they
can combine functionality. The new IC buzzword is System On Chip
(SOC). I started my thesis using a DSP chip from TI that costs over
$100 a piece, ran at 66Mhz. I did some FFT code on the chip and was
very aggrivated with the operation. I then said, hmmmm, let me take
this same code that TI wrote for their chip, modify it (it was in C
and ASM) and use it on a DOS system. The computer was a Celeron 300A
(that I paid $36 from a company on the web). The Celeron outperformed
by an incredible amount.

I don't know if youall are aware but the first PDA running linux is
due out within the month. Its from agendacomputing.com, is $150 and
has LinuxVR on it. It has a bash shell and even Xwindows!! Also with
the uClinux project I think you are going to start seeing linux as a
BIG choice for embedded systems (especially over WindowsCE). I think
this is one of the most exciting things about linux right now.
Imagine buying a DSP that you can upload your ladspa plugins too :)

Rick

Friday, December 01, 2000, you wrote:

>>
>> I don't know if you are aware of this, but If I am not mistaken, Emu
>> is owned by Creative Labs anyways.

JT> Yes, very accurate. And E-mu is actually now E-mu/Ensoniq. All three
JT> companies have or had sound card products, and I have a couple of CL/Ensoniq
JT> AudioPCIs here. One in my Linux box and one in the 'doze machine.

JT> I think the merging of these companies is another indicator that the
JT> field is smearing together, with a variety of hybrid products resulting.

>> Long ago when the SB16 was $250
>> and the WaveBlaster was $250 (which I bought both) I requested the
>> documentation from Creative labs and they sent me the proteus user
>> guide.

JT> Wow, that's interesting... maybe I should look into the Soundblaster
JT> Live! a bit deeper. It might be possible to write an app for it that
JT> turns it into a pretty nice Proteus-like sampler/synth. It'd be _really_
JT> nice to have a Linux driver for the E-mu pro sound card, though.

>> Editing effects stacks on a monochromatic LCD screen
>> is crazy, I would rather have knobs!

JT> It depends on the interface. Both Paul Barton-Davis and I have Alesis
JT> Quadraverbs, and speaking for myself only (I don't need to repeat Paul's
JT> words for him here), that interface is very easy to use. Still, I'd
JT> much rather use my 19" color hi-res monitor, mouse and keyboard. But
JT> Alesis went a loooooong way with a monochrome LCD screen! I am not
JT> complaining.

JT> BTW, regarding editing using knobs, the Proteus family modules have
JT> what E-mu calls "quick edit mode" and "deep edit mode", which allows
JT> the realtime controllers to edit parameters just by turning the knob.
JT> It's a lot quicker than stepping through menus with cursor buttons,
JT> then spinning the data entry wheel. Maybe this is what you were
JT> referring to?

>> Once your product becomes software and not hardware you loose an edge.

JT> I wasn't really arguing in favor of E-mu converting to a software product,
JT> but moving some functionality to the computer for a more hybrid system.
JT> They would still be selling hardware.

JT> Think of it, a lot of what an Ultra sampler does (memory management,
JT> filesystem, SCSI/IDE hardware interface, etc.) would be handled a
JT> lot better by the Linux kernel and cheap PC hardware. In fact, they
JT> could come out with a future product based on Linux if they wanted to.
JT> I wonder if that new Mackie hard disk recorder, for example, is running
JT> Linux inside? Why not?

>> why redo something the right way when you can hack up the old
>> technology with a faster processor :) Believe me, I design IC chips
>> and I see this go on all the time!

JT> My XL-1 has a 40MHz Motorola COLDFire processor, obviously to handle
JT> the user interface and such, while the E-mu custom chips are generating
JT> sounds. Meanwhile, a 1GHz system has a lot of computing power, even if
JT> a good percentage of that gets sopped up in overhead. There's a point
JT> where it's cheaper and easier to throw a PC at the problem than buy some
JT> closed and proprietary solution that had a lot of development cost per
JT> customer associated with it. Since you say you design IC chips, maybe
JT> you understand the cost factor in designing and fabbing a new chip.
JT> (I have had some experience in chip design too...) Meanwhile, a 600MHz
JT> Duron is only $60. Intel spends billions on chip development, and the
JT> competition and huge market drive the price down to peanuts.

JT> I sincerely hope that E-mu holds up, because I really like the company.
JT> Yet at the same time I'm wondering, why don't I just start writing sampler
JT> software for Linux, and in a few years a 4 GHz Linux system might
JT> easily blow away any sampler that E-mu (or AKAI, Yamaha, et. al.)
JT> have by then. Might even be better than Gigasampler. :)

JT> - Jay Ts


New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Fri Dec 01 2000 - 09:44:05 EET