NVIDIA driven by relentless GPU innovation

The latest recommended read from me is the Latest Epic story from NetApp discussing the adoption of  NetApp Storage by NVIDIA. NVIDIA, as you probably know, are leaders in graphics processing and pioneers in the original inventors of the Graphics Processing Unit (GPU). For NVIDIA the requirement to innovate GPU technology is constant and we have seen the application of the GPU technology address many new markets over recent years.


Why is this ? Surely a GPU has one application - accelerating graphics processing capability and performance ? Wrong ! The NVIDIA GPU technology has seen adoption over a  broad range of application areas. Areas such as research, PC gaming, scientific, automotive, supercomputing and of course visualisation. The reason that the GPU as a technology has found new markets is down to its design differences when compared with traditional CPU technologies. While the CPU has seen significant increases in the number of cores introduced on a single silicon wafer, the architecture of the GPU  delivers hundreds of fully programmable cores.


This makes the GPU ideally suited to graphical processing, but also a perfect candidate for High Performance Computing where the need is typically to perform the same calculation at many thousands of data points concurrently. You begin to see why the GPU grew up in the graphics world with all those pixels to manage. One clear application is my old favorite climate and weather modelling as it requires the equations that describe the physics of atmosphere to be applied to millions of data points in 4 dimensions. As you can imagine the processing required to simulate something as complex as the climate over time requires highly complex physics and mathematics and the climate modellers were early to recognise the potential of the GPU as  an engine that could help deliver results quicker.


For NVIDIA the challenge is accelerating time to market. Designing GPUs is a highly complex and time consuming process. Like the climate modellers, new designs for GPUs are modeled and simulated on computer a long before any silicon is etched. The process of design and  modelling also produces plenty of data. In nine months the NVIDIA engineers generated 2.4 billion files and amassed 15 Petabytes of engineering data. But their existing storage was not up to the job. Each day storage problems caused productivity issues which left engineers stranded and frustrated.


NVIDIA recognised that impact the existing storage was having on the business and having evaluated several providers chose NetApp. The NetApp solution based on the FAS6290 and FAS6280 including FlashCache and key OnCommand and Data ONTAP functions such as Deduplication, SnapMirror and SnapVault changed the productivity of their engineers overnight virtually eliminating compute downtime, giving uninterrupted access back to the engineers. The change in productivity also enabled NVIDIA to process a staggering 4.5million jobs per day and doubled the I/O capacity.


So next time you turn on your PC there is a good chance you will see the NVIDIA logo on the startup splash screen, stop for a moment and consider the impact of the GPU technology and what it takes to bring great products to market.


If you would like to read the story yourself then please follow this link:


Follow me on Twitter