Re: [S] Regarding S-PLUS 5.0 Performance

Diego Kuonen (Diego.Kuonen@epfl.ch)
Thu, 03 Dec 1998 15:23:56 +0100


S-Plus users,

reading Andrew Bruce's posting and earlier announcements
made by Mathsoft on Splus 5.0 performances, I could not
believe what William Shannon reported...

> fun <- function()
> {
> apply( array(rnorm(9*10000),dim=c(3,3,10000)), 3, solve)
> }
>
> The 5.0 requires 26 minutes to invert 10,000 3x3 matrices versus 3
> minutes with 3.4.

First, I run the code on a SGI (IRIX Release 6.2 + 1024 Mbytes)
using S-Plus 3.4 Release 1 :

...>unix.time(fun())
[1] 141.27 5.05 150.00 0.00 0.00

..>unix.time(fun())
[1] 142.68 4.92 151.00 0.00 0.00

Second, I did the same on a Sun ULTRA 60 ( SunOS 5.6 + 1024 Mbytes)
using S-Plus 5.0 Release 2:

...>unix.time(fun())
[1] 743.07 2.42 758.00 0.00 0.00

...>unix.time(fun())
[1] 741.79 2.55 752.00 0.00 0.00

There is definitely a problem...

But, why does this happen?

Shouldn't it be the reverse?

:-)

Diego Kuonen

P.S.: Thanks to William Shannon for this benchmark comparison...

--------------------------------------------------------------
Diego Kuonen
Department of Mathematics, EPFL, 1015 Lausanne, Switzerland
Phone: + 41 (0)21 693 5508 Fax : + 41 (0)21 693 4250
mailto:Diego.Kuonen@epfl.ch http://dmawww.epfl.ch/~kuonen

-----------------------------------------------------------------------
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