Archives


- Beowulf
- Beowulf Announce
- Scyld-users
- Beowulf on Debian

[Beowulf] A Mosix Cluster or Beowulf

Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.

Search

Pere Urbón Bayes purbon at deic.uab.es
Thu Feb 23 03:17:54 PST 2006


Mark Hahn wrote:
>>Hello to everybody, I have on mind to build an open mosix cluster with a
>>2 UCP from AMD (DualCore) machines.
> 
> 
> why openmosix (rather than, say, beowulf)?

Sorry! When I mean openmosix i was speaking about to implement a mosix
cluster, and with beowulf a not mosix cluster ( like with MPI, etc... ).
The better subject could be: Mosix Cluster VS MPI (etc, ... ) Clusters..


>>Who's better, AMD or Intel dual-core ucp?.
> 
> 
> it's largely a matter of taste, though there are some less 
> critical differences that matter more to some people.

But I think that AMD Opteron Dual-Core have independent data buses for
the diferents cores, isn't it?

 >>How about the temperature?
> 
> 
> Intel, in general, dissipates more for a comparable machine than AMD.
> this is modulated somewhat depending on power-saving features, 
> but those depend on the machine being idle (which, presumably, 
> anyone in HPC wants to avoid...)

This is what I thought, but the performace of AMD Dual-Core will be
better than the intel.

>>Which do you prefer? How about the performance on a mosix cluster?
> 
> 
> as far as I know, Mosix is still most relevant for non-IO/IPC-intensive
> serial (maybe threaded) jobs. 

This is what i need and for this i think that finally i will use an
openmosix.

 is there an MPI implementation that behaves
> well on Mosix?  it would be sort of interesting to see how a sysv-shm
> MPI behaved on Mosix (since it provides some aspect of network-shared mem).
> that would seem a more promising direction than trying to make socket-based
> MPI migratable...
> 
> but I have no direct experience with Mosix.
> 

Regards

-- 
|---------------------------------------------------------------.--.
|Pere Urbón Bayes      JabberID: purbon at jabber.org             |o_o |
|Dept. of Information and Communication's Engineering          |:_/ |
|E.T.S. d'Enginyeria. Autonomous University of Barcelona      //   \ \
|08193 Bellaterra, Cerdanyola del Valles (Catalonia)         (|     | )
|Tel: +34935813573 QC/2005 (Combinatoric and Coding Group)  /'\_   _/`\
|-----------------------------------------------------------\___)=(___/



More information about the Beowulf mailing list