Last week they were announcing the first Fermi performance numbers. This week, they’re announcing their new CUDA Training Program. EMPhotonics, proprietors of the CUDA-accelerated linear algebra package CULA, have been developing production-quality, commercial CUDA code for some time. Recently, they started offering training courses in stealth mode to customers who were interested in developing CUDA-enabled applications. The feedback from the initial courses was so positive that they decided to begin offering courses to a wider audience.
EM Photonics’ extensive CUDA experience really gave us a kick-start into understanding the CUDA programming paradigm and how to apply this to accelerating real-time signal processing applications. The training material was well prepared and backed up by a very enthusiastic and knowledgeable trainer. I highly recommend for those needing to get rapidly up to speed with CUDA and GPGPU programming,” said David Tetley, Software Engineer Manager at GE Intelligent Platforms.
It was an excellent course on GPU computing and CUDA,” said Dr. C.J. Reddy, President and CTO at Applied EM. “We very much appreciate the level of expertise shared by the EM Photonics engineers and are happy that we spent the time so productively,” he added.
The modular structure of the courses is a bit different than what you’ll find elsewhere. EMPhotonics specifically designed the course curriculum to allow participants the ability to tailor their training sessions for their specific applications and skill level. The on-site training class allows customers to use their own tools to learn in an environment they are comfortable in and the ability to discuss their needs privately.
The feedback we have received so far on our GPU training sessions is overwhelmingly positive. There is a large demand in the market today for CUDA training classes that address specific applications from image and signal processing to computational analysis on platforms from field-deployed systems to workstations and large high-performance computers,” said Eric Kelmelis, CEO of EM Photonics. “Educating others on the techniques EM Photonics has helped pioneer grows the market for GPU-based computing and raises awareness for the platform.”
The two-day standard training program covers parallel computing basics, CUDA programming, cross-platform development, debugging, deployment concerns, GPU optimization, and other topics. Additional time can be added for more in depth instruction on particular areas or to address domain-specific techniques. If you’re interested in more info on the course, check out the EMPhotonics training website here.