SRC® Oil & Gas, Seismic Processing Markets -- Reverse Time Migration

Markets & Applications

Oil & Gas / Seismic Processing

Overview & Challenges

Oil and gas are among the world's most important resources. The petroleum industry is a key driver of the global economy as petroleum is used for numerous commercial and consumer products in addition to serving as the world's primary fuel source. To locate and expand precious resources, the oil and gas industry relies on seismic exploration.

Seismic processing is very computationally demanding, and the volume of data being manipulated is enormous. These advanced imaging techniques often require hours to days of compute time. The industry is continually evaluating new imaging techniques and algorithms for complex sub-surface structures. Petroleum companies must balance the need for rapid subsurface images, with the economics of cost of hardware (number of compute nodes), architectural constraints (bandwidth latency for example) with more mundane but equally important considerations such as power (availability and cost), floor space, and cooling.

Click here to download the SRC^® industry brief on the oil and gas market.

SRC Solutions
Ease of Programming with the Carte^™ Programming Environment
Performance Gains
Related White Papers
Find Out More

Other Markets:

o Defense/National Security

o Quantitative Finance

o Unmanned Aerial Vehicles (UAVs)

o Medical Imaging

o Scientific & Academic Research

SRC^® SOLUTIONS
SRC Computers delivers high performance reconfigurable computer systems that have very high I/O bandwidths to match the orders of magnitude computational speedup that customers in the seismic processing industry require. Attributes of SRC MAPstation^™ workstations and Scalable Systems & Servers that help achieve significant application performance gains include: MAP^® processors, the SRC reconfigurable compute element, that deliver orders of magnitude speedup over microprocessors using very low power and generating very little heat Rotating Common Memory (RCM) that delivers the 3D seismic data to compute at a sustained rate of 3.6 GBytes/sec Global Common Memories (GCM) that can be dedicated to a MAP delivering 3.6 GBytes/sec. Example use: o Reverse Time Migration can use multiple GCMs as input/output data volumes to the MAP processors and ping-ponged between time steps Application specific memory access that does not suffer from cache miss penalties seen by microprocessors. These data transfers can deliver at least 2.8 GBytes/sec bandwidth. Example access patterns: o Matrix transpose o 2D planes out of 3D volumes o 3D mini-volumes o Processor-to-processor communication bandwidths of 3.6 GBytes/sec Learn more about SRC MAPstation Workstations Learn more about SRC Scalable Systems & Servers
EASE OF PROGRAMMING WITH THE CARTE^™ PROGRAMMING ENVIRONMENT
The SRC Carte Programming Environment takes high-level language C or Fortran code, compiles portions of it to run on the implicitly controlled microprocessors and creates the configuration information needed for the explicitly controlled reconfigurable MAP^® processors. Everything needed to control both types of processors is then combined by the Carte Programming Environment into a single Unified Executable. Carte software tools support code development and execution on the hardware, as well as in emulation and simulation environments. Learn more about the Carte Programming Environment
Performance Gains

SRC Application Results

The following table illustrates the performance advantage of a single SRC Series H MAP^® processor compared with a highly tuned code running on a standard microprocessor. The performance gains are achieved by the ability to implement a custom mix of functions for the subroutine.

Speedup numbers below include all overhead, including data movement. All data is for a single MAP processor or a single microprocessor core and assumed 100% scalability for the microprocessor cores. Comparisons of the MAP processor to actual microprocessor based systems would result in even higher speedups due to less than 100% scalability in multicore microprocessor systems.

APPLICATION	MAP PERFORMANCE	SPEEDUP: MAP PROCESSOR VS. STANDARD mP
Reverse Time Migration	13.3 nsec per output migration point	25x* (see note below)

* Speedup relative to a 3.0 GHz Xeon

Note: The performance gain for the Reverse Time Migration application using the Series H MAP processor was dependent upon the size of the 3D seismic volumes. For smaller volumes (300x300x300), the Series H MAP processor performed at least 25 times faster than Intel 3.0 GHz Xeon processors. On larger volumes, the Series H MAP processor preformed in excess of 25 times faster.

Related White Papers*
Reverse Time Migration Performance on the SRC^® Series H MAP^® Processor Reverse Time Migration is the most important seismic data processing method used to recover subsurface images of the Earth's interior. This paper describes the evaluation of a compute intensive two-way finite-difference acoustic wave-propagation kernel used in the Reverse Time Migration technique and resulting significant performance gain over traditional microprocessors. IMPLICIT+EXPLICIT™ Architecture This paper explains the innovative SRC IMPLICIT+EXPLICIT Architecture, which fully integrates Dense Logic Device (DLD) technology and reconfigurable Direct Execution Logic (DEL) with the Carte Programming Environment, delivering orders of magnitude increases in performance. * Please e-mail marketing@srccomputers with your contact information to obtain copies of the papers listed here
Find Out More
Contact SRC Computers today to find out how you can get more performance per watt over traditional microprocessor-based systems. Call (719) 262-0213 or e-mail sales@srccomputers.com to speak with our applications experts.