utk username password

[Daniel Gates]

Greetings Barb and Brian,

I got a username and password set up from utk, do I need to email the cluster administrator to get cluster access?

Thanks!

-Dan

[B.Banbury]

Great news!  I think you should be able to log in to request space here: https://newton.utk.edu/manager/

He will set you up with a cluster account, which will be the same login and password as your UTK ones.  

Barb

______________________
Barb Banbury
Postdoctoral Research Associate
Ecology and Evolutionary Biology
University of Tennessee, Knoxville

[dga...@huskers.unl.edu]

Hi Barb,

I think I got the password set up I'm just waiting to hear back for access. Is the omearalab10 computer connected? I'm getting a connection timed out and I've tried it from a couple of different computers

-Dan

---

From: treevo-...@googlegroups.com [treevo-...@googlegroups.com] on behalf of B.Banbury [darwin...@gmail.com]
Sent: Monday, June 18, 2012 5:34 PM
To: treevo-...@googlegroups.com
Subject: Re: utk username password

[Brian O'Meara]

It had an ethernet cable that was loose. No matter how hard we try to be virtual, the physical stuff keeps cropping up. I've swapped out the cable. 

btw, from your work on profiling I heard that S4 was taking a lot of time, but that fixing it would be beyond scope for your project (which makes sense). In the main trunk of TreEvo we've converted the S4 classes to S3. With just simulation alone, this speeds up TreEvo 3 to 4-fold. So, thank you for finding the S4 issue! We still need to check that it works well for PRC, but then one thing you folks might want to consider is rolling this S4->S3 change into the branch code. It may make profiling easier and allow you to find other places to speed up the code.

Best,

Brian
_______________________________________
Brian O'Meara
Assistant Professor
Dept. of Ecology & Evolutionary Biology
U. of Tennessee, Knoxville
http://www.brianomeara.info

Students wanted: Applications due Dec. 15, annually
Postdoc collaborators wanted: Check NIMBioS' website
Calendar: http://www.brianomeara.info/calendars/omeara

[Derrick Zwickl]

Hi Brian,

Wow, that's great that you were able to do some S4->S3 conversion and gain so much!  When you are sure that the changes don't have side effects, we will definitely want to merge those over to Dan's branch.  Let us know.

Cheers,
Derrick

[Hilmar Lapp]

This makes me curious, too. Brian - for my edification, can you point to some example places in the code where going from S4 to S3 seems to have contributed to this speedup? There may be some useful lessons to be learned in here.

-hilmar

-- 

===========================================================

: Hilmar Lapp  -:- Durham, NC -:- informatics.nescent.org :

===========================================================

[Brian O'Meara]

We just profiled the overall change once we got rid of S3, not the change by steps (it's easiest as an all or nothing thing). But this was after modifying just abctaxon from S4 to S3, and that was a very simple object: just a list of five things (id, name, states, nextstates, and timeSinceSpeciation). All of the S4 calls were calls to read or set one of those values -- I suspect that it's much slower to find the right method to access "id" and then set it, after first making sure you're passing it a number, rather than just doing x$id<-newval. There were no S4 operators on the object itself (unlike abcparticle, which can mutate itself).  

Apparently it's known that S4 is slower, based on a little searching, but it's not really emphasized. Here's a blog post about it http://brokeringclosure.wordpress.com/2008/05/23/s3-vs-s4-efficiency-issues/ .

Brian

_______________________________________
Brian O'Meara
Assistant Professor
Dept. of Ecology & Evolutionary Biology
U. of Tennessee, Knoxville
http://www.brianomeara.info

Students wanted: Applications due Dec. 15, annually
Postdoc collaborators wanted: Check NIMBioS' website
Calendar: http://www.brianomeara.info/calendars/omeara