IA-64 client?

[MAD-ness]

Can't get to the main page at the moment to check, but I don't remember seeing anything previously.

Is there an IA-64 client?

[Aegion]

I took a look, and there certainly does not appear to be one at the moment. I'm thinking that with Intel's involvment in the project it should be possible to get their assistance in making a port to the IA-64 instruction set. Howard, what are your thoughts on this matter?

[MAD-ness]

I imagine they are somewhere along the lines of "#*$& like I don't have enough to do already!"

What his *response* will be, I don't know.

[Brian the Fist]

Ok, everyone who has an Itanium please take two steps forward.

As for Intel's 'support', you'll notice that's been gone for awhile already, if it was ever there..

[pointwood]

Quote (posted by guzzler) from the Ars DC Arcana:

Has anyone tried running the client on IA64 machines? I have several sitting around the office currently not doing anything.

That's the reason for the request

It's what we need to pull ahead of Free-DC in the daylies

[MAD-ness]

Not sure why anyone would have an Itanium or whatever the new ones are called, but if they did it seems odd to have them idle. Oh well, who knows.

I don't imagine there is a ton of demand yet.

[Kapoor]

I have access to three HP i2000 workstations. Two are dual-processor 800 MHz Itaniums the third is a single processor 733 MHz Itanium. For most of the year they have been crunching seti work units using the i386 version of the seti client in Windows XP 64-bit Edition. The i386 seti client ran in the Itaniums x86 emulation mode. This configuration completed a single work unit in just over 25 hours. The dual-procs would average just under 2 work units per day.

Several weeks ago I noticed an IA-64 HP/UX client and decided to give that a try. I "installed" HP/UX and the client and processing times dropped to 2 1/2 hours per work unit. The dual-procs crunch about 20 work units a day.

I emphasized "installed" because my background is Windows and thus my comfort level with any flavor of UNIX is low. I managed to get only enough of HP/UX installed to get the seti client running.

Lately I've begun to look around for another project to get involved with. And any project with a native IA-64 Windows XP command line client would be at the top of the list.

[MAD-ness]

Holy crap. Those SETI numbers are scary.

[Kapoor]

Which ones?

25hrs/wu (i386-winnt) or 2.5hrs/wu (ia64-hpux)? It seems to me that depending on one's perspective either could scary. :-)

[FEEDB0B0]

Hi Folks

I haven't posted much here in a while, but I wanted to let you all know I'm very much enjoying all your posts, suggestions, comments and enthusiasm for our project! I expect the CASP meeting to be exciting and I can't wait to find out how we, collectively, did.

Officially, I should have new computer systems arriving in mid-late October into my lab which will allow us to make the following executables available:

32-bit AIX
64-bit AIX
IA-64 Linux
IA-64 Windows


These platforms are being added to our compile farm thanks to a contribution that IBM is making to our Biomolecular Interaction Network Database project that will also extend our cross-compilation ability for the Distributed Folding project.

Here's a blurb about the BIND project from Genome Canada, one of our sources of research funding for the project. On this page there is a link to a Discovery Channel (.ca) video bit they did about proteins earlier this summer :
http://www.genomecanada.ca/projects/...?id=c2p69&l=e#

In addition I also now have a budget for upgrading our YAC cluster, which will be done by the first quarter of 2003, or perhaps, earlier.

Finally, let me say that Howard and I have worked out a next-generation protein-folding algorithm that I hope we can get massaged into the client by early next year.


Cheers
Christopher Hogue, Ph.D.
Scientist, Bioinformatics
Samuel Lunenfeld Research Institute and
Assistant Professor, Dept. of Biochemistry, University of Toronto

[MAD-ness]

Awesome.

Thanks for the additional info.

One of my teammates (the one producing 10-11 MILLION structures/day - - he is sort of hard to miss) had inquired about IA-64 since he apparently has a few servers sitting around with "nothing to do."

We know you guys are busy (especially Howard, we really do work him to death) but it is GREAT to hear from project management every so often to find out new information about the project, the science and about how well things are going.

Now, back to business. What method would you guys prefer suggestions or requests for client improvements (assuming you re-work it to use a new super-duper algorithm) to be submitted with?

The client is "pretty darn good" as it is, but there are additional features which could make it more appealing (or feasible) to both large farmers and less computer literate users (the key point here being the ability to run the client on a lot more machines and increasing the available computing power by a good bit).

[pointwood]

So, what is the status of those new client versions?

[Brian the Fist]

Heh-heh. Unfortunately plans can change. We didn't end up getting those boxes, ergo no ports. If and when we do get such machines, ports will follow shortly.