Lawrence Berkeley National Laboratory masthead A-Z Index Berkeley Lab masthead U.S. Department of Energy logo Phone Book Jobs Search
Tech Transfer
Licensing Interest Form Receive Customized Tech Alerts

Green Wave: Energy-Efficient HPC Design Optimized for Seismic Imaging
IB-3089, JIB-3153, IB-3162


High Performance Computing (HPC) for



A Berkeley Lab team led by John Shalf and David Donofrio developed Green Wave, a energy-efficient computing platform that can perform critical Reverse Time Migration (RTM) functions at a fraction of the power required of current systems.

The Green Wave is a complete System-on-Chip (SoC) design that uses arrays of energy efficient Tensilica computing cores able to outperform conventional microprocessor and GPU clusters on delivered performance per watt. Key innovations include correct balancing of compute capability with memory caches and local stores; custom extensions to the Tensilica instruction set to accelerate RTM computing performance; and extensions to optimize data movement and communication performance, which makes more effective use of precious memory bandwidth.

The Green Wave design also achieves efficiency by including on a single chip essential resources (the interconnect, memory controllers, and storage controllers) required for HPC. Meanwhile, it dispenses with features such as bloated instruction sets and sound cards that consume power on general-purpose processors but have no use in HPC. The entire node fits onto a chip instead of the complex motherboards required for conventional x86 servers. The result is an energy-sparing architecture optimized for RTM calculations.

Power consumption is today’s limiting factor for HPC performance. Green Wave’s improved energy efficiency translates into reduced operating costs for HPC data centers. In benchmark comparisons, Green Wave demonstrated an 11-fold improvement in RTM code performance for seismic imaging applications over that of conventional (Intel) x86 clusters, and a 3.5x improvement over Graphics Processing Unit (GPU) clusters. The Green Wave ASIC is tailored for HPC applications such as RTM that involve “stencil” computations on block-structured grids, but can be applied to a much broader range of related scientific applications such as CFD and image analysis. This 3D imaging based on seismic data is critical for accurate modeling of geologic formations harboring natural gas and petroleum reservoirs.

Massively parallel supercomputers are essential for demanding HPC applications such as climate modeling, fluid dynamics, and seismic imaging for oil exploration. Power consumption for these HPC systems is hugely expensive, particularly for highly compute-intensive and data-intensive operations such as RTM, which is used extensively in seismic imaging. Because seismic survey ships have insufficient electric power capacity for such systems, computational modeling must be performed on-shore, making real time analysis impossible.

The lower energy consumption afforded by Green Wave opens up new possibilities for real-time HPC. Ship-borne platforms offering real time seismic imaging will be become feasible with Green Wave’s low power profile. Its reduced operating costs will be attractive for other scientific and engineering applications currently burdened by high power consumption. Green Wave can be the launching pad for a growing assortment of application-optimized HPC tasks involving stencil computations on block-structured grids.

DEVELOPMENT STAGE: Bench-scale prototype.

STATUS: Patent pending. Available for licensing or collaborative research.


Krueger, J., Donofrio, D., Shalf, J., Mohiyuddin, M., Williams, S., Oliker, L., Pfreundt, F.J., “Hardware/Software Co-design for Energy-Efficient Seismic Modeling,” SC ’11 Proceedings, Article No. 73, 2011.

Donofrio, D., Oliker, L., Shalf, J., Wehner, M., Rowen, C., Krueger, J., Kamil, S., Mohiyuddin, M., “Energy-Efficient Computing for Extreme-Scale Science,” Computer, Vol.42, No. 11, 62-71, 2009.

REFERENCE NUMBER: IB-3089, JIB-3153, IB-3162

See More Enviromental Technologies