NVIDIA Maxwell vs Pascal processors: Battle of the GPU architectures
NVIDIA's Pascal lineup of graphics processors are expected to be released next year. The company's current desktop, laptop, workstations and mobile SoCs comprise up of the Maxwell architecture and although it is far superior compared to the firm's now obsolete Kepler process, the Pascal GPUs are expected to raise the bar much higher than what consumers would expect.
According to NVIDIA blog, the company's CEO and co-founder Jen-Hsun Huang had revealed details of the upcoming architecture. The executive stated:
"It will benefit from a billion dollars worth of refinement because of R&D done over the last three years."
According to the blog post, two notable differences will be present between the Pascal and Maxwell architecture. The Pascal lineup of graphic cards will be able to accommodate 32 GB of video memory, while the highest amount that the Maxwell process is able to fit in is 12 GB GDDR5, which is the amount present in the company's GeForce Titan X and Quadro M6000 graphic processors.
Additionally, where PC gamers could only place four graphic cards in a motherboard for a quad-SLI configuration, Pascal GPUs will be removing this limitation and allowing users to use 8 GPUs simultaneously for augmented gaming and video rendering performance. A new feature will be added to Pascal's lineup, and it is currently not present in Maxwell. It is called NVIDIA's NVLink, and it is capable of allowing data to move much faster across CPUs and GPUs.
The blog post states that data will be moving between five to twelve times faster compared to the speed delivered by PCI-Express. NVLink is also the reason that gamers will be able to accommodate twice the maximum number of graphic cards in their systems. According to the information present on the blog post's image, it states that Pascal will provide 10 times the performance of Maxwell and while the smaller lithographic process of the Pascal architecture will be a significant perk, additional features stuffed in to the upcoming architecture will also boost performance in upcoming products.
