NVIDIA Reveals Product Roadmaps for GPUs and Tegra SoCs
During his keynote address at GTC 2013, NVIDIA CEO Jen-Hsun Huang revealed the roadmaps for its graphics processors and Tegra chips. Both roadmaps introduced the successors that will replace the incumbents Kepler and Tegra 4 respectively. Read on to uncover more exciting details.
By Wong Chung Wee -
During his keynote address at GTC 2013, NVIDIA CEO Jen-Hsun Huang revealed the roadmaps for its graphics processors and Tegra chips. Both roadmaps introduced the successors that will replace the incumbents Kepler and Tegra 4 respectively.
(Image Source: NVIDIA)
The next generation Maxwell GPU will feature the new "unified virtual memory" where this shared memory space is accessed by both GPU and CPU. Volta, the successor to Maxwell, will feature a groundbreaking technology called "Stacked DRAM" where the video memory chips co-exist on the same silicon strata as the Volta GPU to increase its memory bandwidth to 1TB/s. As a point of reference, the memory bandwidth of the GeForce Titan GPU is 288.4 GB/s while the GeForce GTX 690 has a collective memory bandwidth of 384.4 GB/s (192.2 GB/s per GPU). According to CEO Jen-Hsun, holes had to be drilled through the DRAM layers and allowing them to be stacked together on the same silicon strata as the Volta GPU.
(Image Source: NVIDIA)
The next roadmap revealed was Tegra's and hot on the heels of Tegra 4 are Logan and Parker respectively. Logan is Tegra's 4 successor and it sports a Kepler GPU with support for OpenGL 4.3. It is slated for an official announcement this year and by early next year, it will be scheduled for volume production.
(Image Source: NVIDIA)
CEO Jen-Hsun whetted the appetites of Tegra fans with a side story in the form of developers' kit called Kayla. This platform is powered by a Tegra quad-core ARM processor and a Kepler GPU with support for both CUDA 5 and OpenGL 4.3. It provides an ideal platform for developers to test their ARM-optimized applications, giving them an early start before the official launch of Logan.
(Image Source: NVIDIA)
Parker is the successor to Logan and it will sport a Denver CPU with a Maxwell-based graphics core. It will also feature FinFET transistor architecture that is touted to conserve power as well as the advantage of a faster-to-market delivery that will reduce its risk of obsolescence. On a lighter note, CEO Jen-Hsun introduced Kayla as Logan's girlfriend. This is definitely a delight to early adopters of NVIDIA first CUDA-enabled Tegra SoCs. Stay tuned for more details as we uncover more about these new products in the horizon.
Our articles may contain affiliate links. If you buy through these links, we may earn a small commission.