Haskell GHCi (interpreted) timings:
Integer (arbitrary precision arithmetic)
*Main> foldl1' (+) $ replicate 100000000 52
5200000000
(12.82 secs, 7447947224 bytes)
fl. p. arithmetic:
*Main> foldl1' (+) $ replicate 100000000 52.0
5.2e9
(4.37 secs, 4015930492 bytes)
For a comparison. Here are the J results on my computer:
5 ts '+/100000000#52'
0.999619 536872256
ts '+/100000000#52x'
115.643758 536872768
=@@i
Oleg Kobchenko schreef:
> Difference in speed was kind of expected--maybe not to that degree.
>
> Does it show that modern hardware does not provide enough to scale
> for this kind of calculation?
> Can anyone supply a comparison from other languages or libs: python, Haskel,
> Maple etc?
>
> Interesting was almost the same space.
> I thought it'd blow out the memory.
>
>
> On Sep 27, 2008, at 14:48, Roger Hui <[EMAIL PROTECTED]> wrote:
>
> 52+52 is one hardware instruction and
> 52x+52x is implemented by software,
> hence the 2 orders of magnitude difference
> in time.
>
>
>
> ----- Original Message -----
> From: Oleg Kobchenko <[EMAIL PROTECTED]>
> Date: Saturday, September 27, 2008 10:04
> Subject: Re: [Jgeneral] Limit error
> To: General forum <[email protected]>
>
> Interesting to compare
>
> +/100000000#52 NB. about 1 sec
> 5.2e9
>
> +/100000000#52x NB. about 1 min
> 5200000000
>
> about the same memory.
>
> Let's verify precisely:
>
> 10 ts '+/100000000#52'
> 0.834533 5.36872e8
>
> 2 ts '+/100000000#52x'
> 54.2623 5.36873e8
>
>
>
>
> From: Joey K Tuttle <[EMAIL PROTECTED]>
>
> I thought displaying it would be boring - but maybe not any
> more
> boring than $ or +/ ...
>
> At 08:40 -0700 2008/09/27, Roger Hui wrote:
> The difference is that Joey's expression did not
> try to generate the display of 1e8#52 but Paul's
> expression did, and it's in the generation of
> that display that the limit error was signaled.
>
>
>
> ----- Original Message -----
> From: Joey K Tuttle
> Date: Saturday, September 27, 2008 8:22
> Subject: Re: [Jgeneral] Limit error
> To: General forum
>
> Works OK on my Mac...
>
> $100000000 # 52
> 100000000
>
> version ''
> Installer: j602a_mac_powerpc.dmg
> Engine: j602/2008-03-03/16:45
> Library: 6.02.023ProductName: Mac
> OS X
> ProductVersion: 10.5.5
>
>
> At 15:52 +0100 2008/09/27, Paul Moore wrote:
> >In J32 on Windows, 100000000 # 52 gives a limit
> error. What
> does this
> >mean? I can't see anything stating that there is a
> limit in
> array size
> >(other than available memory) in the documentation.
> >
> >Is it possible for me to create a hundred-million
> entry array? (Such
> >an array of integers would be 400M in size, I have
> 2G so memory
> should>not be an issue).
>
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
>
>
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
>
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
