[14615] in Athena Bugs
Re: sun4 8.0I: athena 8.0 and sparc classics
daemon@ATHENA.MIT.EDU (Greg Hudson)
Fri Aug 30 04:08:30 1996
Date: Fri, 30 Aug 1996 04:08:07 -0400
From: Greg Hudson <ghudson@MIT.EDU>
To: biomorph@MIT.EDU
Cc: bugs@MIT.EDU
In-Reply-To: "[14598] in Athena Bugs"
> use arrow keys and the 'backspace' key on a sparc classic
> they don't do anything. this is in both athena tcsh and emacs
> 19. i'm currently using the 'del' key instead of the 'backspace' key
> for the relevant function
Well, I've looked over your dotfiles, and I can't find anything which
would explain this behavior. Did you solve the problem on your own?
I can try testing with a copy of (the readable parts of) your home
directory at some point.
Are programs finding their app-defaults files okay, or are they
complaining?
On to the more interesting stuff:
> last i checked, sun's workstation prices weren't all that stellar,
> especially bearing in mind that both their hardware and their
> software is generally of the lowest common denominator.
Actually, Sun has very good prices on their workstations, if you take
into account the educational discount. The exact price we get is
confidential, but it's not at all bad.
And, although I'm a free software bigot myself, I consider Solaris a
cut above the operating systems most vendors ship. I've had an awful
lot of bad experiences with AIX, HPUX, Ultrix, and Irix.
> for instance, the dec 5k/133's, in floating point benchmarks i ran,
> were marginally outperformed by the indy's and the sparc 5's. what's
> up with that?
Decstations are a dead product line. You can't buy them any more.
> the snakes that are in the 6.001 lab (hp9000/700's, i believe) blow
> anything athena has out of the water, and they're roughly as old as
> the dec 5k's... why aren't those in the clusters?
They cost a lot of money. You have to compare machines of roughly
equal price if you're going to rate performance; it's meaningless to
consider a donated high-end HP machine to what we buy for the
clusters.
> further, the RS6k/320's were, at the time of their removal, the most
> powerful machine's athena had
Perhaps. For general use, they didn't have enough memory to avoid
swapping and they never worked very well. AIX is a very odd bird, and
very difficult to support.
> now it's true that the indy's have some use. i can concieve of them
> filling some nitch in the ecosystem of academic computing... but to
> populate the clusters with them to the level i see now? and do we
> really need this many microphones and cameras?
A lot of faculty members want us to have a lot of them. I don't like
them either, and they're frightfully expensive. But we can't justify
every decision on the basis of what makes our jobs easier; the faculty
thinks they need SGIs to effectively do academic computing, and that
creates enough pressure that we have a lot of them.
The microphones and cameras are part of the package.
> athena could invest in a butt-load of 486's running
> linux/netBSD/Hurd/freeBSD/minix/xenix-for-all-i-care and populate
> the clusters with those, and even those will punch holes in these
> sparc classics (at least)
Keep in mind that Classics are getting to be old machines now.
We looked into this and decided that no PC operating system would make
a good general cluster platform. Support is always questionable and
no Intel Unix platform has very much commercial application support.
"Not worth the headache." They also wouldn't save all that much
money per seat.
> and sipb already has athena ported to linux and netBSD! and hp-ux,
> last time i checked. though it's a little flaky.
I know; I did a big chunk of the work in all three cases. But porting
the Athena software is a very small part of the effort it takes to get
things running on a given platform; creating the public workstation
platform (installation, update, cleanup, integration with the boot
system, etc.) takes a lot longer.
--Greg Hudson
Release engineer
