> "Small efficient fast-running code" isn't the only issue. A case in point
> in CMU ECE is Matlab: the code isn't that bad, but the data sets get *very*
> large. Large enough, in fact, that we're impatiently waiting for a released
> kernel with large swap partition support ("released" because we're unlikely
> to get a stable AFS client for 2.0.36pre-patches :-) for the folks who want
> to run their jobs without having to fight for access to the Suns and Alphas.
>
> And before CMU I was consulting for a bank department which runs financial
> simulations a lot. One simulation they were experimenting with used 2GB of
> RAM and an equal amount of swap....
For banks this is a small server. Not rarely they're already waiting for
bigger iron than for example Sun's Starfire aka E10000 which has a maximum
configuration of 64 CPUs / 64gb memory. Banks are currently buying these
boxes like normal people buy milk.
I expect that's going to be some problem for the Linux comunity as well. It's
not going to be easy to provide an OS that both fits a 386SX/8mb and top of
the line iron of Starfires or Origin 2000 caliber - and there were already
first steps of porting Linux to that class of machines.
Ralf
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/