[Lowfer] WOLF development

[David Willmore]

> Peter Barick said:
> I am not using WOLF and likely will not be, but that is not due to
> hardware constraints. However I applaud you interest in the current LF
> user population by asking what level processor they are running.
> 
> If the software can be packaged for different levels processor
> configurations, ie, speeds, without compromising the whole, great, as
> long as users know or can find out what level they should employ the
> system at. Hopefully the end result of your changes is a more inclusive
> application and especially for those like me with a older but trusty
> Pentium.

Hey, just doing what Stewart Nelson charged me with doing.  He's let 
me play with his WOLF code.  I mentioned to him how nice it would be
to take advantage of modern processor features (SSE, SSE2, 3DNow, MMX)
to make the code faster.  He cautioned me to not *rely* on any 
characteristic of the machine.  That goes for memory, etc.  So, I
thought I'd poll the group and see what we had to work with instead
of coding for the worst.

It looks like it will be easily possible to take advantage of the
advanced math features in a 'if it has it, use it' fashon.  Memory
is normally harder to deal with as most code is designed to use
X and always X MB.  I think I see a way to do a size/speed trade
off.

That would allow people who have PIII/800's and .5 GiB of memory
to fly through decodes, while still allowing 486DX4/125s w/32 MiB
the ability to decode *at all*.  Yeah, it's going to be slower
and, maybe, even less accurate, but you don't have to throw it
out.  Best of all, it doesn't cust the fast/big people anything.
Sure it costs me a few hours worth of coding and testing, but
that's free, right?  ;)

In another email Les Rayburn spoke thusly:
> I think it would be more constructive to have this discussion on list...
> so that others can flag problem areas. I currently run WOLF on
> two computers.
> 
> 1. A dual 800mhz Pentium III with 512 Megs of RAM.
> 2. A 1.6ghz Athlon with 512 Megs of RAM.
> 
> Both run Windows 98 Second Edition.

I was only intending to take the poll on "what hardware do you have?"
off list as it gets silly any tedious to see a couple dozen emails
listing processor and memory capabilities.  Once I have a firm idea
on what I have to play with, I'll bring the discussion back on list
to see what the group things we should do with what we have.  If 
my idea at a size/speed trade off doesn't work, then we may have to
decide to leave the 486's behind.  Keep in mind that the current 
code/executable will still work.  All these people will lose is
new features, not old ones.  Ones that can easily be ported back
into the old code base they could get, as well.

There is no current plan to become incompatable with the existing
on air nature of WOLF.  That might change, but such a decision
will be made with the full input of the group.

> Personally, I think it's a flaw of most ham radio ops to expect software
> to continue to run and perform well on DOS, 486 machines, etc.

I agree, but it is a hobby and we show a degree of creativity by
finding uses for what others find as junk.  Think of how the US
for the VHF bands, for example.  We got them because they were
unusable.  I mean, they're so high, who could ever make a transimitter
for them? ;)

> Cheap is one thing. Impractical is another. Machines with tons of
> performance can be picked up for less than $500 brand new these
> days. Decent used computers can be had for much less...

Yep, but we'll have friends in contries where that may not be true
and I'm sure a good number of them will be awsome DX. :)

> If the whole point of WOLF is pushing the envelope of weak signal
> detection, then I don't think that worrying about exceeding the ability
> of CPU's should even be an issue. If you're serious about weak signal
> work, then get a serious computer.

Point taken.  To be honest, the buy with the 486DX/4-125 is a
far cry from where the rest of the group clusters.  Your #1 system
is typical for what most people have--it may be their faster
system, but they have one comparable to it.  There are even nuts
that make your #2 system look small.  *shudder*

My own systems--that I'll be using for development are much a good
deal smaller than the average size the group is using.  I use a
PII/400 laptop to do my development and I'll do some performance
tuning on my 800 MHz durron and 433MHz Alpha 21164A--just for grins.

Oh, to ammend my first post, the WOLF code runs just fine under
Linux, windows is *not* a requirement.  If it runs under Linux,
there's no reason it shouldn't run on any other UNIX.  Does that
make anyone other than me happy?

> Boy, was that a soapbox issue or what? I'll just step down now and
> go stand quietly in the back....

Hey, it's getting to be the off season.  'tis time to discuss
now that the time for action has passed.  Speak away!

I'll get back to the group this friday with the poll results and,
maybe, the results from the size/speed trade off option analysis.

Till then, please keep the votes coming in.  There have to be
more people out there.  You don't even have to *be* running WOLF,
maybe just *interested* in *maybe* running it.

Think about what you want to run next winter.  Will you be running
it then?  Well, maybe you'll be running whatever Stewart's new
program is, but, for now, ignore the man behind the curtan!

Cheers,
Daivd N0YMV