|
Authored by: jesse on Tuesday, December 18 2012 @ 08:11 AM EST |
And that cluster provided a "single system" image to the users. As I
remember reading, doing a "ps" to see the process status would display
processes from across the cluster. This would have been around 1995-96(the
article, the cluster was in 1994)
In (or around anyway) 1998, I was working at a supercomputer center and was
present at an IBM SP-4 presentation (they were doing preliminary work to get one
installed). I had just seen a Linux boot log from IBM in Germany - It was
obviously an SP node... with about 60+ disk drives attached.
so scaling was already done, and made the "not scale" statements
stupid.
Not to mention the Fujitsu Sparc system with 20 processors in an SMP
configuration. The limit at the time was 32 processors - as the kernel used a
bitmap to help manage those processors (an unsigned 32 bit integer for speed).
And windows was having trouble with only 4.[ Reply to This | Parent | # ]
|
|
|
|
|