s-news
[Top] [All Lists]

[no subject]

To: PConnolly@grunt.marc.cri.nz
From: ripley@stats.ox.ac.uk (Prof Brian Ripley)
Date: Tue, 31 Mar 1998 08:09:33 +0100 (BST)
Cc: S-News@wubios.wustl.edu
In-reply-to: <199803310206.OAA29579@grunt.marc.cri.nz> from "Patrick Connolly" at Mar 31, 98 02:06:35 pm
Sender: owner-s-news@wubios.wustl.edu
Patrick Connolly wrote:
> 
> 
>  Barry Courtois wrote:
> 
> |> Rather than given an overall impression of which particular implentation
> |> of S you think is faster than another, I would find it more helpful if you
> |> gave concrete examples.  Perhaps a judicious use of dos.time() and
> |> unix.time()?
> 
> That sounds fair enough and I would have done it except that
> dos.time() returns a value of something like 4.8 seconds, but it
> doesn't include the 10 minutes I was waiting for the hourglass to
> finish.  Running the same code in Unix would have a unix.time of
> something like 7 seconds.  Comparing the "objectively measured" times
> does not give a sensible comparison.  If I can see that I can start
> doing something else in Unix in less than 10 seconds, but need more
> than 10 minutes in NT, I believe my subjective measurement and
> distrust what was being measured by dos.time().

On my 133Mhz Pentium laptop, W95, 32Mb of memory, S-PLUS 4.x (all
varieties) starts up in 40-50 seconds.  S-PLUS 3.3 starts in 7 seconds.
S-PLUS 3.4 on my Sun UltraSparc 1/170 also starts in 7 seconds. (On our
1990-vintage diskless 20Mhz Sun SLCs it takes about 30 seconds depending
on the network load.) On our 200Mhz Pentium, 64Mb, NT4.0 S-PLUS 4.0 also
takes 40 seconds. (In all cases, starting in an empty directory with no
object browser.)  Sqpe.exe starts up in a few seconds, about 2 seconds the 
second and subsequent times it is used (due to caching).

> I brought up this question last year and it didn't seem to be common
> to others.  There's a bottleneck somewhere that doesn't affect Word
> and Excel and I don't really blame S4 for it.  I haven't had the time
> to seek it out thoroughly.  I need to borrow other equipment to test a
> few things to track it down.

The bottleneck seems to be the very large number of files (DLLs and start-up 
files) that S-PLUS 4.x processes before getting to the prompt.


-- 
Brian D. Ripley,                  ripley@stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu.  To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message:  unsubscribe s-news

<Prev in Thread] Current Thread [Next in Thread>