Re: [q-lang-users] Re: Optimizations and tweaks

[Tim H. <q...@st...>]

Dr....@t-... (Albert Graef) writes:

> Tim Haynes wrote:
>> If it's not too late, I'd like to hold back on that - I'm travelling atm
>> and will be back home tomorrowish. Can run yours and mine through the mill
>> again then. I'm particularly worried about those error-in-conditional
>> problems... :/
>
> Hi, Tim,
>
> unfortunately Q 6.0 was already released last weekend (didn't have the
> time to announce this on the ML this time, I'm so busy now that the
> semester has begun). 

That is not unfortunate ... :)

> My corrected version of your script is included in the examples. That
> versions seems to work ok, though. I'd suggest that if you have a new
> version I'll update it in cvs immediately and then release it with the
> next Q version. Ok?

I've just got some initial benchmark results - it's still good, it's
still obviously quicker than the naiive baseline (faster about 3.5x as
often as naiive is faster than mine). And when it's faster, it's by a
factor of 2-3, while I've never seen it being over 12% slower at worst.

Graphs of times taken show a couple of curves much lower than the naiive
algorithm, too.

Neat :)

~Tim
-- 
<http://spodzone.org.uk/>