[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Best,Fastest platform for IDL 5.2 (NT or UNIX)
- Subject: Re: Best,Fastest platform for IDL 5.2 (NT or UNIX)
- From: rivers(at)cars3.uchicago.edu (Mark Rivers)
- Date: Tue, 14 Sep 1999 03:57:06 GMT
- Newsgroups: comp.lang.idl-pvwave
- Organization: CARS, U. of Chicago, Chicago IL 60637
- References: <37DD70F9.D76719BB@sbrc.umanitoba.ca>
- Reply-To: rivers(at)cars3.uchicago.edu
- Sender: news(at)midway.uchicago.edu (News Administrator)
- Xref: news.doit.wisc.edu comp.lang.idl-pvwave:16524
In article <37DD70F9.D76719BB@sbrc.umanitoba.ca>, Richard Tyc
>To add to the another similar post, I would like to know from the
>experts what system they would buy if they had $10K - $30K (speed being
>an important factor) ?
>We are running on a SGI O2 R5K, 576Mb RAM and it is pathetically
>slow!! Our project has recently received some infusion of capital and I
>would like to ask what hardware platform would be ideal ? I am not
>opposed to switching over to NT.
>The application makes heavy use of object graphics, volume rendering
>with cutting planes etc. (eg. render volumes of 512x512x100 with
>real-time motion updates using the trackball object)
>I was thinking of moving up to a SGI Octane with the R12K CPU (or
>Any performance comparisons with IDL on the new Pentium III 600 MHz
>machines vs. UNIX workstations ?
I am using a Dell Precision 610 workstation with 1GB of RAM, 36 GB of RAID 0
disk, and dual 450 MHz processors. I am very pleased with the system. Its
cost today (with 550 MHz processors) is about $8,000.
My application is 3-D tomography data processing and visualization. Having 2
processors helps in 2 ways:
- I can be running one IDL session doing compute-intensive reconstructions on
one processor while doing interactive work on the other.
- The IDL volume rendering object does use both processors
1 GB of RAM is essential for what I am doing, in fact I still need more memory
at times, but Windows has a limit of 1 GB of virtual memory per process. The
memory fragmentation problems often seen in IDL under Unix are much less
serious under Windows. I can create and delete 300MB arrays interactively for
hours without having to restart IDL.
I have not used similarly configured Unix systems so I really can't compare.