DSP programming in Atari days

четверг, 6 сентября 2018 г., 15:18:44 UTC+6 пользователь Randy Yates написал:
> versa20@gmail.com writes:
> 
> > понедельник, 3 сентября 2018 г., 5:48:31 UTC+6 пользователь Randy Yates написал:
> >> versa20@gmail.com writes:
> >> > [...]
> >> > Thank you! I see these books you mentioned is for understanding of
> >> > DSP. But what resources did you use to learn programming? Do manual
> >> > were distributed with computers? How these manuals were useful for
> >> > learning programming?
> >> 
> >> I think I used K&R a good bit at first. Of course you need the
> >> compiler/linker manual, which for me I believe was a Microsoft compiler.
> >> The graphical bit was machine-specific but I was on a VGA device.
> >> 
> >> However, I suspect something is woefully wrong with your mental model of
> >> the process. I spent 90 percent of my time learning the theory.
> >> 10 percent was used to program. Programming the algorithms is almost
> >> trivial (at least for me).
> >> -- 
> >> Randy Yates, DSP/Embedded Firmware Developer
> >> Digital Signal Labs
> >> http://www.digitalsignallabs.com
> >
> > Programming side seems most complicated for me than learning DSP
> > theory. Rigth now I've got info about which DSP book I need to read
> > and now it looks less problem than "How to write working code?" type
> > of problem.
> > I can get all theory (formulas, algorithms, etc) from these books, but I don't know how to make software from it by using 68k Assembly
> 
> Try opening that DSP book and reading it. With understanding. Work the
> problems. 
> 
> See if you have the mathematical foundations on which DSP is based, like
> algebra, calculus, complex variables, probability theory, etc. Make sure
> you know Fourier transforms, Laplace transforms, Discrete-Time Fourier
> Transforms, the DFT, z-transforms, the Goertzel algorithm, etc. Some
> knowledge of sequences and series would also be helpful. Oh yeah, you
> know how to handle Dirac delta functions?
> 
> What you seem to think is going to take, oh, a few days, takes most
> engineering students (including myself in my younger days) years of
> study to get through.
> 
> But perhaps you are a genius... 
> 
> And I'm with Eric - can you tell us why you're hung up on 68k assembly?
> You do realize how old that processor is, right (30-40 years)? Most
> processors within the last 20 years would kick the crap out of a
> 68k with sheer MIPS, L1/L2 cache, etc. There are also newer processors
> made for the purpose such as TI's TMS320C55x, 6xxx, etc. 
> -- 
> Randy Yates, DSP/Embedded Firmware Developer
> Digital Signal Labs
> http://www.digitalsignallabs.com

I'm learning C language as you adviced for me. Probably I'm gonna to leave that idea to program DSP for Atari ST, but I still interesting how Stephan Bernsee implemented morphing algorithm for Atari ST1040 in the late 80s. I would like to replicate that achievement and evolve my knowlegde and skills of programming. Also other programmers advised me to use the emulators of older platforms to improve my programming skills.

On Thursday, September 6, 2018 at 5:22:34 AM UTC-5, ver...@gmail.com wrote:

> I'm learning C language as you adviced for me. Probably I'm gonna to leave that idea to program DSP for Atari ST, but I still interesting how Stephan Bernsee implemented morphing algorithm for Atari ST1040 in the late 80s. I would like to replicate that achievement and evolve my knowlegde and skills of programming. Also other programmers advised me to use the emulators of older platforms to improve my programming skills.

The old ways were born of necessity, not because they were the best ways. I programmed an IMSAI 8080 with front panel switches because I had to. The only thing I learned from it was that it was very inefficient. I assert that there are better ways now.

<versa20@gmail.com> wrote:

>I'm learning C language as you adviced for me. Probably I'm gonna to
>leave that idea to program DSP for Atari ST, but I still interesting how
>Stephan Bernsee implemented morphing algorithm for Atari ST1040 in the
>late 80s. I would like to replicate that achievement and evolve my
>knowlegde and skills of programming. 

WAG (wild-ass guess): you want to be a Chiptune musician?

Steve

On Monday, September 3, 2018 at 4:39:50 PM UTC+12, Steve Pope wrote:
> <versa20@gmail.com> wrote:
> 
> >Programming side seems most complicated for me than learning DSP theory.
> >Rigth now I've got info about which DSP book I need to read and now it
> >looks less problem than "How to write working code?" type of problem.
> >I can get all theory (formulas, algorithms, etc) from these books, but I
> >don't know how to make software from it by using 68k Assembly
> 
> You do not need to write 68K assembly, but as assembly languages go
> it's a pretty straightforward experience.
> 
> I think you should start with K&R C.  I was given two weeks to
> learn C, whereas I've spent decades learning DSP algorithms.
> 
> If your target is a 68K -- I've used them extensively -- you
> will have to write maybe a hundred lines of assembly to create
> a system that then runs code that is compiled from C.  Easy-peasy.
> 
> But today, I'm not sure why anyone would use a 68K instead of
> say a Raspberry Pi.  (Maybe some low-power requirement? Or simply
> nostalgia?)
> 
> Steve

The level of programming knowledge for DSP is very limited. Mostly multiply and add. This is not general purpose programming or Java etc which is far more complicated. Unless you make a fancy user interface of course with fancy graphics etc. Most real-time programming is just a while loop and do something within it. Exception I suppose is handling interrupts and I2C sensors, writing drivers. Even then its understanding teh data sheet more than the actual programming that is needed.

Am Mittwoch, 29. August 2018 12:19:29 UTC+2 schrieb Gerhard Hoffmann:
> Am 29.08.2018 um 06:02 schrieb Steve Pope:
> 
> > 
> > That wasn't what it was like for me.
> > 
> > The go-to component .. if your project could afford it, was the
> > TRW 16-bit multiplier (later, multiplier-accumulator).  Another popular
> > component was the 2901 bit-slice.  Most requirements could be met
> > by surrounding one or both of these with appropriate circuitry.
> > 
> > Not much DSP was done on microprocessors, but projects did include
> > devices such as 1802's or 6809's.  Intel processors were avoided due
> > to their jittery execution times.
> > 
> > Steve
> 
> 
> That's how I did a FIR filter in 1984:
> 
> < 
> http://www.hoffmann-hochfrequenz.de/project_gallery/project_gallery.html 
>                >

I like the drill in your lab!

 
> (I admit, my web site needs an update desperately.)
Haha :-)