Picasso is a scientific computing cluster funded in part by a generous donation from the Intel Education Initiative. It is used to solve complex problems encountered by researchers working at the ECE department in the University of Texas at Austin. This is a heterogeneous supercomputer with a peak performance of 1.8 TFlops jointly used by Profs. Yilmaz, Erez, Chiou, Caramanis and Viswanath.


Architecture: Picasso consists of 1 head node and 44 compute nodes. Each node has one 2.66 GHz Intel Xeon 3220 quad-core processor for a total of 180 cores. Picasso compute nodes are configured with 352 GB of total memory and 7.04 TB of local disk space (8 GB memory and 160 GB local disk on each compute node). The maximum number of processors that can be used is 176.


Interconnect: The connections between the compute nodes and the frontend are through an InfiniBand switch as well as a Gigabit Ethernet switch. All nodes are connected through the ethernet switch and Compute-0-0 to compute-0-22 are also connected through the IB switch.


Selected Presentations

All Publications

Publications by Topic




Prospective Students