The submissions for the Analytics and StorCloud Challenges are done. Analytics is our weakest link, if you will forgive me, because we had to cobble it all together - from start to completion - here and now rather than in time for SC05 like the other two. The abstracts for Analytics and StorCloud are done. The extended abstract and summary are also done for Analytics. We have an abstract, extended abstract, network diagram, and resource requirements due by the end of the month. After doing the Analytics, that ought to be a piece of cake.
We have all our vendors onboard. AVS is willing to give us a test license for their parallel AVS product. SGI is willing to lend us a half a rack of an Altix (20 +/- Itanium/2s) for the visualization in the booth at SC05. We're using an IBM SP here at NERSC. We're also using our embarassingly parallel cluster (PDSF) here at NERSC for the data reduction. DDN is providing us with 30 terabytes of disk on one of their dual controller storage boxes. Pogo Linux is providing the GPFS storage nodes. IBM's GPFS is how we're tying this all together and we have the test license to let us do just that.
We're going to be running CAM which is a part of CCSM 3.0 on the SP and writing to the GPFS file system. The data is then reduced on PDSF and visualized on the Altix in the booth tied to the 50 in plasma display there too.
I'm the one tying this all together. I have a backup, but that's if Avrora, my wife, or I get sick only. Otherwise everything system wise is on my shoulders. Fortunately, we're gonna have test shots first here at NERSC to make sure it all works ahead of time.
Cross your fingers!!!
We have all our vendors onboard. AVS is willing to give us a test license for their parallel AVS product. SGI is willing to lend us a half a rack of an Altix (20 +/- Itanium/2s) for the visualization in the booth at SC05. We're using an IBM SP here at NERSC. We're also using our embarassingly parallel cluster (PDSF) here at NERSC for the data reduction. DDN is providing us with 30 terabytes of disk on one of their dual controller storage boxes. Pogo Linux is providing the GPFS storage nodes. IBM's GPFS is how we're tying this all together and we have the test license to let us do just that.
We're going to be running CAM which is a part of CCSM 3.0 on the SP and writing to the GPFS file system. The data is then reduced on PDSF and visualized on the Altix in the booth tied to the 50 in plasma display there too.
I'm the one tying this all together. I have a backup, but that's if Avrora, my wife, or I get sick only. Otherwise everything system wise is on my shoulders. Fortunately, we're gonna have test shots first here at NERSC to make sure it all works ahead of time.
Cross your fingers!!!
No comments:
Post a Comment