Jaguar is a Cray XT3 provided as a primary system in the National Leadership Computing Facility (NLCF).
The current Jaguar installation has 5294 nodes, each with a 2.4-GHz AMD Opteron processor and 2 GB of memory. 5212 nodes are available in the compute partition, and the remainder provide I/O and login services.
Each node is connected to a Cray Seastar router through Hypertransport, and the Seastars are all interconnected in a 3D-torus topology. The resulting interconnect has very high bandwidth, low latency, and extreme scalability.
The operating system is UNICOS/lc, which is a combination of Linux on the service nodes and the Catamount microkernel on the compute nodes. Catamount is designed to minimize system overhead, thus allowing scalable low-latency global communication.
Phoenix is a Cray X1E provided as a primary system in the National Center for Computational Sciences (NCCS).
Phoenix has 1024 multi-streaming vector processors (MSPs), where each MSP has 2 MB of cache and a peak computation rate of 18 GF. Four MSPs form a node with 8 Gb of shared memory. Memory bandwidth is very high, roughly half the cache bandwidth. The interconnect functions as an extension of the memory system, offering each node direct access to memory on other nodes at high bandwidth and low latency.
The Cray X1E uses custom-designed vector processors to get high performance for scientific codes. The Cray designed processors are linked together using a high-performance shared memory interconnect technology. Phoenix has 1024 multistreaming vector processors (MSPs), each of which can carry out as many as 18 billion operations per second, making the performance of the total system as high as 18.5 trillion operations per second.
