An overview of the BlueGene/L Supercomputer
Proceedings of the 2002 ACM/IEEE conference on Supercomputing
High-density computing: a 240-processor Beowulf in one cubic meter
Proceedings of the 2002 ACM/IEEE conference on Supercomputing
MegaProto: 1 TFlops/10kW Rack Is Feasible Even with Only Commodity Technology
SC '05 Proceedings of the 2005 ACM/IEEE conference on Supercomputing
Power and environment aware control of Beowulf clusters
Cluster Computing
Megaproto/E: power-aware high-performance cluster with commodity technology
IPDPS'06 Proceedings of the 20th international conference on Parallel and distributed processing
| Hi-index | 0.00 |
"MegaProto" is a proof-of-concept prototype for our project "Mega-Scale Computing Based on Low-Power Technology and Workload Modeling", implementing our key idea that a million-scale parallel system should be built with densely mounted low-power commodity processors. It also serves as a platformto implement and evaluate our new technologies such as power conscious compilation, highly reliable and high performance networking, highly dependable cluster management, and multi-level scalable parallel programming. The building block of the MegaProto is a 1U-high 19 inch-rack mountable motherboard unit on which 16 low-power, one-dollar note-sized, commodity PCarchitecture daughterboards are mounted with a high bandwidth, 2Gbps per processor network based on Gigabit Ethernet. The peak performance of each unit is 14.4GFlops for the first version and will improve to 38.4GFlops in the second version through a processor/daughterboard upgrade. The intra- and inter-unit network bandwidths are 32Gbps and 16Gbps respectively. As for power consumption, the entire unit idles at less than 150W and consumes 300-330W maximum under extreme computational stress; this is comparable to or better than conventional 1U servers comprised of dual high-performance, power hungry processors, while benchmarks exhibit up to 279% superior performance for some NPB programs. This demonstrates that higher performance can be achieved with low-power, densely populated architectures with commodity components.