Acceleware at RICE Oil & Gas HPC Workshop in Houston

Rice University, Houston, Texas was the venue of this year’s Oil and Gas High Performance Computing Workshop held February 28. A record number of attendees met to discuss, share and learn about the latest technology advancements and future industry challenges. Dirk Smit, Vice President Exploration Technology and Chief Scientist Geophysics at Shell, started off the workshop with an interesting keynote on the compute challenges of future energy demand, which paved the way for a variety of other speakers and presentations.

Scott Quiring, Software Developer and Seismic Team Lead at Acceleware, joined industry and academic leaders to present a case study on ‘Reverse Time Migration: A Study on GPU/CPU Hybrid Computing’ as part of the Accelerator Enabled Computing parallel session.

Scott’s presentation demonstrated the importance for RTM implementations to be continuously re-balanced as hardware performance and software features evolve over time in order to avoid performance bottlenecks and maintain optimum throughput. Other key points of his presentation included:

  • The basic methodology of RTM
  • RTM computational challenges including efficient propagation of a 3D seismic wavefield and accessing the wavefield in reverse-time order
  • How to maximise performance on hybrid (GPU/CPU) platforms

A recording of Scott’s presentation can be viewed here:


Does your CUDA application need to target a specific GPU? If you are writing GPU enabled code, you would typically use a device query to select the desired GPUs. However, a quick and easy solution for testing is to use the environment variable CUDA_VISIBLE_DEVICES to restrict the devices that your CUDA application sees. This can be useful if you are attempting to share resources on a node or you want your GPU enabled executable to target a specific GPU.

Vegas, baby, Vegas!

The Society of Exploration Geophysicists (SEG) headed back to Las Vegas for their International Exposition and 82nd Annual Meeting. An impressive 8,412 delegates attended the event making the show the world's largest oil, energy and mineral exposition.

New in the Acceleware booth for 2012 were dynamic talks on the latest research developments for FWI, 3D elastic modeling, RTM and hardware acceleration advancements. Acceleware’s RTM Product Manager, Darren Foltinek and Seismic Researcher Team Lead, Dr. Ray McGarry teamed up with Ty McKercher, Principal Solution Architect from NVIDIA to deliver the presentations:

  • An Introduction to Full Waveform Inversion
  • AxRTM: An Innovative Reverse Time Migration Library
  • Beyond RTM, GPU Integration in the E&P Workflow
  • Evaluating Hardware Acceleration for your Workflow
  • Making 3D Elastic Modeling a Reality
  • Dispersion Reducing Finite Difference Coefficients for CSEM

If you missed the presentations or you would like a copy of the slides, drop us an email at

An Introduction to FWI
An Introduction to FWI

SuperComputing 2012 Wrap up

We set off to Salt Lake City for SuperComputing 2012 (SC12), trying to escape the snow and cold in Calgary only to land in a full blizzard at SLC airport. The snowfall didn't last very long though and we experienced the city in a beautiful mix of fall, winter and blue skies.

Acceleware - Salt Lake City by Night

First C++ AMP Course in Seattle, WA

Last week I had the pleasure of teaching (drum roll please!) the first EVER C++ AMP class. The course was held on Microsoft’s headquarters in Redmond, WA, just down the hall from the development team that created this new technology that is bringing GPU programming to the masses.

Acceleware C++ AMP Training Course
I knew I had found the right building at Microsoft’s headquarters in Redmond, WA when the lamp posts lining the street were all decorated with the new Visual Studio 2012 logo!

C++ Accelerated Massive Parallelism (C++ AMP) is a new open specification heterogeneous programming model, which builds on the established C++ language.

Created for heterogeneous computing C++ AMP is designed to accelerate the execution of your C++ code by taking advantage of data-parallel hardware such as a GPU, with an implicit fallback to multi-core CPU and SIMD instructions.

Developed in partnership with Microsoft, the course focused on using GPUs as the data parallel hardware. The 12 attendees were from a range of disciplines including Metro App & Gaming developers to C++ and C# developers looking to accelerate a myriad of algorithms. The C++ AMP advantage of productivity proved itself throughout the week as students quickly grasped the API, worked with the intuitive and broad functionality of the C++ AMP debugging support in VS 2012 and used functionality, unique to C++ AMP, to optimize their algorithms. For example, the course participants wrote functions with the restrict(amp,cpu) qualifier that allows them to write the function once but call it from both CPU and GPU code. Additionally, students used the new completion_future class to optimize, with clean readable code, asynchronous operations.

Thank you to all of the attendees who participated. I really enjoyed teaching the class and I look forward to the next course in New York!

EAGE 2012 in Copenhagen, Denmark

The 2012 meeting of the European Association of Geoscientists and Engineers took place in Copenhagen, Denmark, a beautiful, clean, friendly city with an outstanding mixture of historical buildings and very modern architecture. We stayed in the old part of the city, close to a canal where ships of all descriptions sail by. The city streets are dominated by folks in business attire riding bicycles to and from work, and the resulting lack of car traffic makes for a quiet and very pedestrian-friendly city core. There is a quick and efficient driver-less train that brought us to and from the conference center which is distinguished by the very modern attached hotel, with it's sloping and overhanging walls and triangular windows. The delicious Danish open sandwiches called smørrebrød (literally, buttered bread), were a big hit with all of us for lunch, as was the local Tuborg beer.

Copenhagen Conference Center - Acceleware at EAGE 2012

We saw a regular stream of visitors and the two RTM images that showcased our latest TTI imaging abilities got a lot of attention, with several people commenting on the excellent quality of the image. We were located close to the Paradigm booth, and had several meetings with our friends there. We have also had several discussions with Tsunami Developments, our other RTM integration partner. Both are very keen to integrate against our next AxRTM release, due out by the end of June.

Reverse time migration is becoming a more established technology this year, with several medium-sized processing companies offering it as a service, in addition to the major acquisition and processing companies. Likewise, GPUs are a relatively established technology, with all high-performance hardware vendors offering GPU-enabled solutions, and several of the RTM talks being based on GPU implementations. However, most of technology that was presented was either already available in our AxRTM library or things that we are aware of and working on.

Full-waveform inversion forms a large part of this years technical session, with four full sessions and a two-day workshop and course dedicated to the latest research in FWI.

Med venlige hilse fra Danmark,
- Darren, Ray, Steve, Geoff

GTC, San Jose 2012

This year at NVIDIA’s GPU Technology Conference in San Jose, I had the privilege of hosting a CUDA™ workshop to over 200+ attendees. Throughout the day we covered topics including GPU memory management, debugging, and optimizations. The questions from the audience were very insightful and addressed important aspects of GPU programming. I would like to thank Kiran and Ramesh who were able to help me on stage with the code demonstrations. The feedback from attendees was very positive and I look forward to hosting additional seminars at upcoming conferences.

Video: Acceleware Reverse Time Migration

As oil and gas exploration is forced to deal with increasingly complex geologies, many of the established imaging methods (for example Kirchhoff depth) can struggle to produce accurate images. Reverse Time Migration (RTM) has become firmly established in recent years as the preferred imaging option for complex geologies. Its simplicity and superior imaging quality stem from the use of a full solution 2-way wave equation. This makes no approximations limiting the direction in which seismic energy can travel.

Slidecast: GPU Technology and Usage Models

GPUs and their applications can be a complex topic. This week I came across an insideHPC article of NVIDIA presenting an introduction into the GPU technology and usage models. So, if you are interested in an introduction of what GPUs can do for you, look no further!

Webinar: Hardware Accelerated Clusters

Acceleware latest webinar: Watch Acceleware Product Manager Chris Mason discussing the benefits of using GPU clusters for performance and power savings:


Subscribe to RSS - blogs