by Lou CoveyEDITORIAL DIRECTOR
The data industry has been limping along with processing systems designed for personal computing and jury-rigged for big data and high performance computing. It has worked for light-touch computing applications like Facebook, but for compute-intensive applications like the Oculus Rift virtual environment and Oracle's database in-memory, the hardware infrastructure is woefully insufficient (even Oracle's hardware).
The Rube-Goldbergesque contrivances that have brought us this far have made it easy to ignore investment in new approaches, but they will fail us in the foreseeable future and retard growth in data-intensive enterprises. And the efforts of the design automation industry to fix the insufficiency through additional kluges are as unprofitable.
New Tech Press met with the renowned computer scientist John Gustafson who confirmed that the infrastructure cornerstones (the x86 platform of Intel and ARM cores) will not bring us into the new age of high-performance computing, even with help from graphics processors from NVIDIA and AMD. Gustafson posits we need something new, and that investment in that yet-to-be determined new platform needs to happen now.