comp.lang.idl-pvwave archive
Messages from Usenet group comp.lang.idl-pvwave, compiled by Paulo Penteado

Home » Public Forums » archive » Linux Help Needed
Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend 
Return to the default flat view Create a new topic Submit Reply
Re: Linux Help Needed [message #47242 is a reply to message #47153] Fri, 27 January 2006 17:55 Go to previous messageGo to previous message
JD Smith is currently offline  JD Smith
Messages: 850
Registered: December 1999
Senior Member
On Fri, 27 Jan 2006 19:47:53 +0000, Greg Hennessy wrote:

> On 2006-01-27, Nigel Wade <nmw@ion.le.ac.uk> wrote:
>> It seems that the Intel Macs are not all that Apple would like you to believe:
>>
>> http://www.macworld.com/2006/01/features/imaclabtest1/index. php
>
> Not that I have a dog in this fight, since I use Linux on opterons,
> but a different viewpoint can be found at:
>
> http://www.macspeedzone.com/html/hardware/machine/performanc e_in_the_raw/06/1_23.shtml

My strong impression is that, for large arrays (say 200,000 elements
and more), the MacBook should be around 4-5 times faster at basic
processing in IDL (not reading/writing to disk, drawing to screen,
etc, but FFT's, arithmetic, etc.) than the Powerbook G4's they
replace. In fact I'd be willing to make a wager to this effect.

Each core of the Core Duo is about as fast as a G5. If you are
hammering away at both cores, you'll really embarrass your old PB. I
expect *very* similar performance for Linux laptops with the Core Duo
and the MacBook: same compiler, very similar libraries, same
processor, same memory sub-system, etc. Macs may lag a bit due to the
extra overhead of function calls, but it should be close. It will be nice
not to suffer the "PPC" penalty under IDL (though we lament the unrealized
potential which was Altivec).

Macworld's test was something of a joke. Though useful, because it
reflects real world typical-use performance, we could easily design
benchmarks in IDL which actually stress the dual CPU cores to their
maximum, and we would find much different results. Here's a hint. This
is not a CPU benchmark:

IDL> openw,un,/get_lun & writeu,un,lindgen(10000000L) & free_lun,un


JD
[Message index]
 
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Previous Topic: q5/ ***Hot stuff - check this out !!! q5/
Next Topic: Fastest Nearest Neighbor Calculation

-=] Back to Top [=-
[ Syndicate this forum (XML) ] [ RSS ] [ PDF ]

Current Time: Fri Oct 10 04:13:33 PDT 2025

Total time taken to generate the page: 0.96132 seconds