OT Linux (RE: [rescue] OT: Stuffed Proliant?)

[Jochen Kunz]

On 2001.12.22 23:13 Joshua D Boyd wrote:

> The 32bit limit isn't too bad.  For one thing, it is a 32bit per page
> limit. You actually can have 36bits of address space.  So, if you
> partition you DB into 4gig pages, then you can handle 64gig DBs.
1. That is a work around (read poor hack), no solution.
2. In a typical *ix environment the amount of RAM that a single process
can use is lower than the max. virtual address space. The memory map
defines different areas in a address space for shared libs, the
executable, stack, ... How much RAM can be used is highly OS specific.

> What limitations are imposed by DOS 1.0 compatibility?
You have to carry all that legacy stuff like real mode, A20 gate, the
poor interrupt concept, ... with you. This may not be a real limitation,
but it is usless overhead that can cause problems. These are historical
artifacts, that are only present, because nobody tries to do a radical,
state of the art redesign like IBM did with the MCA PS/2 line. (Or Apple
with the PowerMac / PCI PowerMacs to have a look beyond the PeeCee
pond.) "A PeeCee has to be compatible, regardless of the cost."
--



tsch|_,
         Jochen

Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/