RDAV_SAB

sizzlepictureΛογισμικό & κατασκευή λογ/κού

2 Δεκ 2013 (πριν από 3 χρόνια και 8 μήνες)

133 εμφανίσεις


Any opinions, findings, and conclusions or recommendations expressed in
this material are those of the author(s) and do not necessarily reflect the
views of the National Science Foundation

RDAV Update

Sean Ahern


Director of the UT RDAV Center

Science Advisory Board Meeting

19
-
20 July 2010


Executive summary


RDAV resources (Nautilus) are currently in the allocations system, and several
requests have been made.


Largest of which is from Jim Kinter, here at the SAB meeting


Though much of our hardware and software is in place and ready to go,

engineering delays from SGI have pushed our timeline for full deployment out.


This SGI UltraViolet is one of the first off the assembly line.


More I/O bandwidth than any other UltraViolet system ordered (30
-
50 GB/s).


Current full availability timeframe is the middle of August.


We continue to pursue early customer activities and educational opportunities.

Hardware status


Nautilus
: Engineering delays from SGI have pushed our full acceptance date back a
few weeks.


1/4 of the full UltraViolet machine is going through acceptance testing now and seems to
be working well.


The rest of the UltraViolet machine arrives at NICS today (7/18/10) and will be fully
available mid August.


Graphics cards
: NVIDIA has not met their deadlines for full scalability. The
timeframe for 16 Tesla S2050‘s is the fall. We decided to deliver a machine without
the full GPU complement rather than further delay delivery for the NSF.


Four prior generation NVIDIA GPUs will be available in the interim.


Parallel filesystem
: We expect to have a full 1 PB parallel filesystem available on
Nautilus on August 1.


Portal
: Our portal system is operational and we are working to deploy new
capabilities on it.

Software and environment status


VisIt has been ported and runs well on our smaller system.


Remote visualization systems are deployed and secure: NX, VNC


Workflow systems appear to be working.


Compiling environments work well, though we’re waiting on SGI for the full
complement (Intel compilers, primarily).


Exploring issues with job placement, NUMA aware scheduling, and GPU scheduling
through our job management system (MOAB/Torque).

RDAV Portal


Completed:


SSH
-
VNC portlet for remote access to RDAV resources tested for functionality.


Integration of access to Dashboard components tested.


Continuing work:


Teragrid authentication


Teragrid queue and account information access


To Do:


RDAV visualization tool launcher


Semantic visualization product store


Workflow manager portlet

RDAV Portal: Main Portal

RDAV Portal: eSimMon
Dashboard

Work with early users


Have staff working directly with three early users to push early system.


Bronson Messer: Supernova simulation:


Worked on efficiently converting large parallel data into format suitable
for exploratory visualization with VisIt.


Scripted batch processing to generate imagery for Bellerophon portal
system.


Stephen Miller: Spallation Neutron Source experiment:


Rewrote rebinning filter to process experimental data from scattering
experiments into a usable format: Datasets from 4 GB
-
several TB.


Wrote custom VisIt scripts to produce visualizations to smooth
workflow


Lou Gross: National Institute for Mathematical and Biological
Synthesis:


Statistical analysis of species diversity in the Great Smoky Mountains
National Park.


Novel visualizations for species presence and absence predictions

Greater Smoky Mountains National park

MaxEnt prediction for Acer Saccharum

Supernova entropy

Education, Outreach, and Training
activities


We taught a joint visualization class with TACC at the

Petascale Programming
Environments and Tools classes on 9 July.


We will be teaching a tutorial on Nautilus usage for visualization, data analysis, and
workflow management at the TeraGrid'10 conference.