- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello.
In our class there are 20 computers, Intel Core 2 Duo in each. They are interconnected by simple 100Mbit Ethernet network using one Ethernet switch and twisted-pair.
The question is: is it reasonable to use ScaLAPACK functions (e.g. pdgetrf) at this network and 20 computers? Or I have to use just LAPACK functions (e.g. dgetrf) at single computer?
Thanks!
In our class there are 20 computers, Intel Core 2 Duo in each. They are interconnected by simple 100Mbit Ethernet network using one Ethernet switch and twisted-pair.
The question is: is it reasonable to use ScaLAPACK functions (e.g. pdgetrf) at this network and 20 computers? Or I have to use just LAPACK functions (e.g. dgetrf) at single computer?
Thanks!
Link Copied
2 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello,
depends the task size. if the task size match with RAM available on single computer then please use Lapack version of dgetrf if not then -ScaLapack.
--Gennady
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thanks for your answer!
Yes, the taks's data can be stored in one RAM on single computer.
Another question - is there any solution (may be not Intel :)) that will yield some performance in such Ethernet network?
Yes, the taks's data can be stored in one RAM on single computer.
Another question - is there any solution (may be not Intel :)) that will yield some performance in such Ethernet network?
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page