[Beowulf] Altix vs. 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.
Joe Landman landman at scalableinformatics.comWed Apr 20 14:43:05 PDT 2005
- Previous message: [Beowulf] Altix vs. Beowulf
- Next message: [Beowulf] Diep speeds at altix versus quad opteron
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Mark Hahn wrote:
>>Please note that several network cards also have build in shared memory at
>>the card. Quadrics has a shmem library for example at a real low cost for 8
>>nodes.
>
>
> a shmem library does not shared memory make.
long *big_honking_storage_pointer,
billions_and_billions = 16 * 1024 * 1024 * 1024;
big_honking_storage_pointer =
malloc(billions_and_billions*sizeof(double));
/*
ahhh global shared memory, tastes great, less filling, easy to
program
*/
Basically if you cannot allocate a huge chunk of ram and have the OS
figure out where to get it from without breaking the request up, then it
really isn't shared memory. If you cannot slide along contiguous
address blocks on a single CPU and touch all of your ram, then it really
isn't shared memory.
There is some sort of Jeff-Foxworthy type thing happening here.
---
Joe
- Previous message: [Beowulf] Altix vs. Beowulf
- Next message: [Beowulf] Diep speeds at altix versus quad opteron
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
