[m-dev.] question about mode representation

[Charles Shuller]

On Fri, Mar 30, 2018 at 1:37 AM, Michael Day <mikeday at yeslogic.com> wrote:

> For all that time, the compiler will need more memory to represent programs
>> than it does right now, and initially, the time needed to fill in the
>> new field won't generate any returns in terms of performance. (There will
>> be
>> such returns later, when switching to use the new field allows a compiler
>> pass to avoid threading the current instmap though.)
>>
>
> Developer time is worth more than memory usage, so whatever approach makes
> software engineering smoother is probably the way to go.
>
>
If the extra memory is only required during compilation, which seems to be
the case, then I agree whole-heartedly, Use the RAM :)



If the extra memory is required in the generated executable, then I think
the question is more complex.

Memory Required maps directly to Monthly VM Cost, so we always consider it
when making decisions about which tech to use.   Additionally, consider
that for embedded development small memory footprints are a hard
requirement.

The extra required memory seems like it might be as much as 1 word per Var,
which in many cases will double the required RAM (Consider a program of
only integer variables) and possibly the required stack space.   Increased
stack space requirements strongly imply function calls will be slower as
well, since you have to push more stuff onto the stack, or if your able to
use registers for function calls, you're more likely to fill up available
registers and start pushing stuff onto the stack.



Cheers!

Charles
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mercurylang.org/archives/developers/attachments/20180330/1b005779/attachment.html>