AMBER Archive (2008)

Subject: Re: AMBER: AMBER 10 PMEMD / SGI_MIPS

From: Atro Tossavainen (atro.tossavainen+amber_at_helsinki.fi)
Date: Tue Dec 16 2008 - 06:17:48 CST

Bob,

> I have actually never indicated support for Sun SPARC in pmemd;

That's right - I was referring to AMBER 10 in general and should have
made the explicit distinction w.r.t. PMEMD vs the rest of AMBER 10.

Am including my sun_sparc.f90.mpi PMEMD (v8) config as attachment in
case you're interested :-) I see you've moved from csh style configs
to sh style in the meantime.

> 2) not having indications that this is a scorching chip I should be
> paying attention to.

LOL! Er... SPARC certainly isn't the hottest number cruncher around,
never was, I think.

It happens that the Finnish centre for scientific computation still has
(for a while, anyway) a fairly large Sun box (dual single-image E25k's,
with a few hundred GB of memory and 96 dual-core UltraSPARC IVs) that
might be more useful for us for AMBER work than stand-alone PCs are
(and is handy to have around when their main Linux cluster fails
catastrophically and is out of use for several days in a row as has
been the case this week).

(The other boxes at CSC are a HP CP4000 BL and a Cray XT4/XT5. Getting
 AMBER and PMEMD to run on the CP4000 was a no-brainer, but I'll probably
 be harassing you in a bit with XT* issues.)

> I am still a bit agog on the namelength thing. In general, I would
> probably prefer to not have to live with such an onerous restriction,
> but I understand it is annoying for you. So if it is just this one
> name, you can of course easily hack up the code to get it to compile.

I don't think I have a way of determining whether it's just this one
name or others too short of trying to compile the program and seeing
all the places where it bombs. Fortunately it was the only name that
caused the problem and I've got what I think should be a working install
of AMBER 10 including MPI versions of sander and pmemd on the SGI now.

> I think the deal with the MIPS architecture is that it is just so slow
> as to be pretty much useless now - basically not worth the power it
> takes to run them. I would think some Origin's might still be running
> out there, though, as it was a pretty nice smp.

Indeed. The 128CPU 160GB 16*Origin 2k here at the Finnish CSC was
dismantled in May 2005 already, though.

> Are we dealing with a couple of workstations at U of Helsinki,

I am dealing with a few single CPU workstations and a 4CPU smp. This
does not count as serious MIPS iron by any standards. I totally agree
it's outdated, but as long as there's the option to configure for this
and I've still got some hardware here I'm going to give it a try every
time I get my hands on a new release of AMBER. :-) 

-- 
Atro Tossavainen (Mr.)               / The Institute of Biotechnology at
Systems Analyst, Techno-Amish &     / the University of Helsinki, Finland,
+358-9-19158939  UNIX Dinosaur     / employs me, but my opinions are my own.
< URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS
----------------------------------------------------------------------- The AMBER Mail Reflector To post, send mail to amber_at_scripps.edu To unsubscribe, send "unsubscribe amber" (in the *body* of the email) to majordomo_at_scripps.edu