Path: csiph.com!eternal-september.org!feeder.eternal-september.org!mx02.eternal-september.org!.POSTED!not-for-mail
From: Ben Bacarisse <ben.usenet@bsb.me.uk>
Newsgroups: comp.lang.javascript
Subject: Re: Improve speed calculations?
Date: Mon, 11 Jan 2016 15:44:21 +0000
Organization: A noiseless patient Spider
Lines: 43
Message-ID: <87fuy4w27e.fsf@bsb.me.uk>
References: <86bb6f49-d74d-426d-bc2a-5b4cd7199bf4@googlegroups.com> <69c96cdd-1020-470c-b0b2-66e2898468bb@googlegroups.com> <87si24w8id.fsf@bsb.me.uk> <5693bf11$0$23861$e4fe514c@news.xs4all.nl>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit
Injection-Info: mx02.eternal-september.org; posting-host="017616aa25f81ec581c44d76d61ba2f3"; logging-data="20196"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1/gy5H3mbHh6gX9EDGv5hIutvwknAAZaB4="
Cancel-Lock: sha1:Myj3rR4qZyBQiBuuszaRAOKn9jA= sha1:RycLYM0A4HDswOW1O6RfNWavY1g=
X-BSB-Auth: 1.a84c22f1a02eed5242f6.20160111154421GMT.87fuy4w27e.fsf@bsb.me.uk
Xref: csiph.com comp.lang.javascript:29230

Silvio <silvio@internet.com> writes:

> On 01/11/2016 02:28 PM, Ben Bacarisse wrote:
>> jonas.thornvall@gmail.com writes:
>>
>>> Den söndag 10 januari 2016 kl. 14:33:27 UTC+1 skrev jonas.t...@gmail.com:
>>>> <script language="Javascript">
>> <snip code>
>>> http://jt.node365.se/BAUTABASE.html
>>>
>>> Using the biggest base possible adding "4503599627370495" encoding
>>> fibonacci in javascript without overflow.
>>>
>>> Also fixed ambiguity reading out the numbers in bases over ten by
>>> using separator for digitplaces. The add can not cause overflow but to
>>> get rid of base overflow would require some real deep thinking.
>>>
>>> The 100000th Fibonacci took 34 seconds i do not think i can get it
>>> much faster in javascript if your a wizard please prove me wrong.
>>
>> I have not tried your code, but I've just written a quick bignum add
>> function and I can get fib(100000) in about 3.5 seconds.  If I have time
>> to unravel yours, I'll do a comparative test.  Note that a language with
>> a good bignum type (Haskell, for example) will do it in less than 100th
>> of a second.
>>
>> Note that your 'challenge' makes it easy to optimise the code.  For
>> example, using a power of 10 as the base makes the conversion for
>> printing very simple.
>>
>
> That is well neigh impossible if you used a 10-based decimal since
> Jonas uses a huge base and is convinced that using a larger base will
> make computations much faster. Your result would invalidate that
> claim...

Using base 10 is quite slow (my code needs 23s for fib(100000) using
base 10), but I use a higher power of 10 as the base for speed whilst
keeping printing simple.  With higher powers, you do get faster results
(up to a limit, of course!).

-- 
Ben.