In general, you do have a valid point, but keep in mind, we are *gearing
towards* machines of a higher class. In doing so, I don't think our
demands on the resources of PCs are that limiting.
Win95 needs 32 Megs to run reasonably at all, and that's just the GUI and
NO servers, FTP, HTTP, none at ALL! Win NT, well, we won't talk about it.
Solaris, I have limited experience, and none on x86, but I'm sure Linux
rocks tha pants off it. What does that leave you with to compare? DOS???
In short, you make an inescapable point, but one I think Linux has adhered
too at least comparably well. (If not miraculously)
-Shawn
<=========== America Held Hostage ===========>
Day 2007 for the poor and the middle class.
Day 2026 for the rich and the dead.
915 days remaining in the Raw Deal.
<============================================>
On Mon, 20 Jul 1998, John Alvord wrote:
> Not really. A sort algorithm doesn't influence operating system code very
> much... and the applications which make heavy use are very complicated.
>
> On a less powerful machine, you can't get away with serious inefficiences
> because things will stop dead. You have to be careful will all algorithms,
> memory usage, scheduling, etc. On a Linus-class machine that is all noise,
> unnoticeable really. The differnence between a 3 minute build and a 2.5
> minute build... who really cares. The machine is probably idle 95% of the
> time anyway.
>
> That same Linus-class machine operating under heavy ISP load does have
> limits, though. And making sure that the les powerful machine runs
> efficiently will *help* ensure that the heavily loaded Linus-class machine
> performs at its best.
>
> This argument is not absolute, of course. And memory is an excellent
> tradeoff for computation and I/O capacity. But would we ever have noticed
> the fragmentation problem if everyone was using 512meg machines? The ISP
> INN monsters would have... that is for sure.
>
> Think of the less powerful machine as the canary in the mine that conks
> out before people do.
>
> john alvord
>
>
>
> On Mon, 20 Jul 1998, Shawn Leas wrote:
>
> >
> > This is a highly ignored point, thanks for making it. John was making
> > dangerous generalisations.
> >
> > I suggest that anyone needing examples go back to school, per say, and
> > compare their sort algorithms.
> >
> > However true this may be, an algorithm can modify it's behavior if, say,
> > it does not have optimal resources. Is this closer to the idea you were
> > trying to portray John?
> >
> > -Shawn
> > <=========== America Held Hostage ===========>
> > Day 2007 for the poor and the middle class.
> > Day 2026 for the rich and the dead.
> > 915 days remaining in the Raw Deal.
> > <============================================>
> >
> > On Mon, 20 Jul 1998, Andrea Arcangeli wrote:
> >
> > > On Mon, 20 Jul 1998, John Alvord wrote:
> > >
> > > >Keeping it running on rusty old PCs has a significant benefit, *if*
> > > >the programming effort isn't too high. When a system runs fine on low end
> > > >machines it will fly on high end machines. Inefficiences that are masked
> > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> > > I mainly agree but it' s not always true. For example there are algorithms
> > > that run in O(n*log(n)) but use the double of memory. These algorithms
> > > could be slower in a machine that will need to swapout everything to run,
> > > but they will be very very faster in a machine with a lot of memory.
> > >
> > > Andrea[s] Arcangeli
> > >
> > >
> > >
> > > -
> > > 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.altern.org/andrebalsa/doc/lkml-faq.html
> > >
> >
>
-
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.altern.org/andrebalsa/doc/lkml-faq.html