Your account already exists. Please login first to continue managing your settings.
The next generation of data center architecture will harness the processing power of GPUs and CPUs.
Inundated by data, the oil and gas industry is desperately in need of a more efficient way to process its data. Large data centers have existed for years, but the recent introduction of GPUs into the processing realm is opening new avenues for more efficient data processing.
Petrobras, for instance, designed the grifo04 supercomputer to facilitate its presalt exploration. Earlier this year, the cluster was named the 68th fastest computer in the world and the fastest in Latin America.
According to Paulo Roberto Souza, HPC consultant for Petrobras, the computer acts as a “critical tool” for subsurface imaging.
“This computer has huge processing power to run algorithms specially developed within the company,” he told Digital Look, an online magazine in Brazil. “Thus, using this tool that combines hardware, software, and high technology, our geophysicists generate more accurate images.”
Souza will be one of the participants in an upcoming Hart Energy webinar titled, “Make more accurate prospect decisions in less time with a hybrid computing environment.” The webinar is the second in a series of three “Big Data and the Cloud” webinars that examine issues around data center architecture. It will be held at 10 a.m. CST Nov. 1.
GPUs are an important ingredient in the data center recipe. The adoption of this technology, along with flexible server and storage virtualization and Cloud computing, is causing companies to rethink their HPC infrastructure, according to a white paper from NetApp.
“Colocating GPUs, CPUs, and data avoids many of the data bandwidth, data management, and other challenges associated with dispersed visualization systems while allowing computing resources to be fully used at all times for a combination of visual and HPC applications,” the paper noted. “The ability to virtualize GPU resources – analogous to the way server virtualization solutions virtualize CPU resources – will make it possible to provide a flexible pool of GPUs that are available for both visualization and computation needs.”
The webinar, sponsored by NetApp, also will feature: Ty McKercher, HPC solution architect for NVIDIA; Melinda McDade, strategic performance engineer for NetApp; and John Thomas, technical solution architect for Cisco’s data center group.
For more information or to register for the webinar, visit https://secure.oilandgasinvestor.com/webinars/?eventid=133&where=E%26P.
Contact the author, Rhonda Duey, at email@example.com.