Re: [avr-libc-dev] Even faster decimal code

[Martin McKee]

When it comes to RAM vs. ROM, I usually have issues with running out of
RAM.  Then again, my applications tend to be very RAM hungry.  I am always
keeping large blocks of data in memory.  When I do more standard interface
applications, I find that I run out of ROM first.  I would say that I
generally find working around lack of RAM easier in my applications.  So,
if I have to hit the wall with one, I prefer it to be RAM.

Martin Jay McKee

On Wed, Dec 28, 2016 at 11:40 AM, George Spelvin <address@hidden>
wrote:

> > I find all this fascinating, but I'm really not the one to be commenting
> on
> > what the best approach here is.  I will say, however, that in many of my
> > applications, I would be more likely to chose a speed increase over
> reduced
> > memory.  I tend to live with mostly compute-bound, control applications
> > though.
>
> Thank you very much, this is very useful!  Even if it's just one person's
> experience, at least it's *a* data point.
>
> May I ask, whay do you think of my RAM-for-ROM tradeoff idea?
> Is there one or the other that you more commonly run out of?
>
> > I've not had enough time to look at the code to make fine-grained
> > suggestions (and I'm out of practice with AVR ASM), but I've not felt
> that
> > the comments were too bad.  I've been able to follow the code as written
> > easily enough.
>
> Yes, my original algorithm was excessively tricky.  The last (and
> fastest) one is actually a lot simpler.
>