Visit additional Tabor Communication Publications
November 26, 2008
The growing adoption of high performance computing on small scale clusters by companies from all segments of the economy is driven by the same forces that led HPC to become an integral part of the fabric of science and engineering decades earlier: HPC helps users get things done they just couldn't do before.
In science and engineering that has meant test flying new aircraft designs years before a prototype could have been built in a traditional physical testing workflow, or rationally searching for new therapeutic compounds based on specific desirable biological properties rather than hit and miss experimentation. These advances have created real gains in the standard of living for most of the people on this planet. This is scientific computing in the large, and the impacts are culture shifting. But although the scale of the computation might be smaller, the shift that adoption of today's HPC technologies is causing for businesses and their customers is nonetheless transformative.
I recently talked with Julien Noel, the CT (computed tomography) Product Manager at North Star Imaging in chilly Rogers, Minnesota. North Star Imaging (NSI) specializes in industrial X-ray for nondestructive testing and analysis. They have seen firsthand how the adoption of HPC -- in the form of expanded computational power from NVIDIA's GPUs and their CUDA API -- can transform a business and create new value for them and for their customers.
HPCwire: What does North Star do for its customers?
Noel: Our 2D digital X-ray systems are often used throughout the manufacturing process for product quality control and manual or automated approval/rejection applications. Our 3D CT systems have typically been used for research and development, failure analysis, reverse engineering and other similar tasks.
Our products and services are geared toward anyone who needs to inspect an object internally and/or externally without destroying it. We are involved with industries such as aerospace, medical device, electronics, automotive, museums and many more, and have had the opportunity to work with companies such as Boeing, Bell Helicopter, Lockheed Martin, NASA, US Army, Medtronic, Kawasaki and the list goes on.
HPCwire: What is the problem you are solving with HPC -- in this case NVIDIA's GPUs?
Noel: Computed tomography involves complex algorithms for 3D reconstruction. Basically, the industrial CT system takes several 2D digital X-ray images and reconstructs them into a 3D volume made of voxels or volume elements. This process uses a filtered back-projection algorithm called the Feldkamp Algorithm.
Due to the improvement of digital X-ray technology, industrial CT systems are able to take more X-ray projections than ever before --from 720 to 3,000-- plus each individual image is becoming appreciably larger. Single images now reach 3 to 10 megapixels and have a bit depth usually around 14 to 16 bits. Overall, CT software manipulates massive datasets, as well as creates and outputs multi-billion voxel-sized reconstructions.
In order to process the data and create the reconstruction, the CT software requires a high-end computer with significant computation capability. To date, the standard has been either a multi-core processor system or a computer farm, which in turn leads to expensive hardware and a limit in reconstruction speed. Basically, the CT reconstruction speed is linear with the number of processors -- that is, 8 cores equals 8 times faster.
To combat this issue, NSI developed a new way to perform reconstruction using GPU technology. GPU reconstruction accelerates the process significantly more than traditional CPU systems and also helps reduce hardware costs. GPU technology is the future in regard to computation limits and is very strategic to NSI's future developments as well.
Our GPU-based CT software additionally includes a 3D rendering tool used to display the results and manipulate the 3D CT volume in real time. NVIDIA graphics cards are also used to optimize the rendering quality and speed.
HPCwire: Why NVIDIA and CUDA?
Noel: Historically, we used NVIDIA graphic cards for the 3D rendering module of our CT software. For our CT reconstruction development, the CUDA interface was so easy to use and so powerful that our team naturally focused on NVIDIA technology, with computation on the Tesla C1060 Computing Processor. Moreover, the NVIDIA staff has been very reactive and helpful in providing our team with technical solutions and graphics cards for evaluation and development.
HPCwire: What improvements have your customers seen, and how has that made a difference?
Noel: Through the use of our new CT software with GPU reconstruction technology, our data transformations are completed significantly faster than they had been. With our efX-CT software, speeds are between 5 and 40 times faster than our CPU version using processors, and up to 100 times faster than any other CT software, depending on the number of projections.
From a customer's perspective, this new capability is extremely valuable as it is now possible to run considerably more CT scans per day than ever before. Also, it allows the flexibility and time to try multiple reconstruction settings to fine tune the scan quality.
Recently, a customer explained that their productivity has dramatically increased due to their reconstruction times diminishing by a factor of 50 over machines built just a couple of years ago. They further explained their data is now typically available in less than two minutes instead of hours and the way they utilize computed tomography has changed dramatically.
HPCwire: What level of effort was required of North Star to take advantage of NVIDIA's GPUs with CUDA?
Noel: North Star Imaging's development team has been very efficient in integrating the GPU reconstruction module in efX-CT. The CUDA programming interface is quite straightforward and very stable. The code is easy to get into and its flexibility and capability makes it possible to have something working pretty quickly. More development effort was needed to optimize speed, but overall, CUDA opens up great development perspectives.
HPCwire: Now that you've been through this project, do you see additional opportunities in other products or areas of your business?
Noel: The GPU capabilities using NVIDIA's supercomputer systems have been a great improvement to our efX-CT software and today, most of NSI's CT system customers are using it. Automatic CT systems with fast GPU reconstruction are currently being developed for inline CT inspection and 3D metrology. Also, we are working on more 3D rendering capabilities for real-time and interactive inspection using NVIDIA products as well as fast data filtering for reconstruction quality improvement.
New NVIDIA products with more cores and enhanced memory will definitely bring our software, and in general our CT business, to a whole new level. The way we see it, 3D CT empowered with HPC technologies such as NVIDIA's GPUS is definitely the future of industrial X-ray.
May 22, 2013 |
At some point in the not-too-distant future, building powerful, miniature computing systems will be considered a hobby for high schoolers, just as robotics or even Lego-building are today. That could be made possible through recent advancements made with the Raspberry Pi computers.
May 16, 2013 |
When it comes to cloud, long distances mean unacceptably high latencies. Researchers from the University of Bonn in Germany examined those latency issues of doing CFD modeling in the cloud by utilizing a common CFD and its utilization in HPC instance types including both CPU and GPU cores of Amazon EC2.
May 15, 2013 |
Supercomputers at the Department of Energy’s National Energy Research Scientific Computing Center (NERSC) have worked on important computational problems such as collapse of the atomic state, the optimization of chemical catalysts, and now modeling popping bubbles.
May 10, 2013 |
Program provides cash awards up to $10,000 for the best open-source end-user applications deployed on 100G network.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/15/2013 | Bull | “50% of HPC users say their largest jobs scale to 120 cores or less.” How about yours? Are your codes ready to take advantage of today’s and tomorrow’s ultra-parallel HPC systems? Download this White Paper by Analysts Intersect360 Research to see what Bull and Intel’s Center for Excellence in Parallel Programming can do for your codes.
In this demonstration of SGI DMF ZeroWatt disk solution, Dr. Eng Lim Goh, SGI CTO, discusses a function of SGI DMF software to reduce costs and power consumption in an exascale (Big Data) storage datacenter.
The Cray CS300-AC cluster supercomputer offers energy efficient, air-cooled design based on modular, industry-standard platforms featuring the latest processor and network technologies and a wide range of datacenter cooling requirements.