Beyond Maxwell: Nvidia Announces Their Next-Generation Pascal GPUs

Beyond Maxwell: Nvidia Announces Their Next-Generation Pascal GPUs

Beyond Maxwell: Nvidia Announces Their Next-Generation Pascal GPUs

This week, Nvidia’s GPU Technology Conference keynote was filled with announcements. In addition to showing off the $3,000 Titan Z, CEO Jen-Hsun Huang also updated Nvidia’s graphics architecture roadmap, taking a first look at the Pascal GPU.

This is the graphics technology that will appear after the Maxwell architecture is finished and is expected to appear in 2016. I’m not quite sure what happened to the Volta GPU architecture that was recently released at CES after Maxwell was released in January, but after Pascal, this seems to have been delayed by at least two years.

The new GPU architecture, code-named after scientist Blaise Pascal, will introduce three-dimensional memory and a new interconnect called NVLink that will provide 5 to 12 times the bandwidth of current PCIe connections.

This 3D memory, also known as stacked memory, means that the VRAM will be stacked directly on top of the GPU itself. “For the first time,” Jen-Hsun said, “we’ll build chips on top of other chips. We’ll be stacking heterogeneous chips—that is, different types of chips—on one wafer.”

They then punched tiny holes in the silicon to allow connections through the DRAM stack all the way to the GPU. “Together it forms an interface that provides an incredible amount of bandwidth,” he continued. “A giant leap.”

Nvidia estimates that we will see 2 to 4 times the memory bandwidth currently available. It also allows for 2.5 times the capacity and an impressive four times the energy efficiency of current GPUs.

The new NVLink interconnect, combined with the new GPU design, greatly improves PCIe bandwidth and increases the available bandwidth for multi-GPU arrays with Pascal GPUs. This should enable more advanced, possibly linear, scaling as you add more and more GPUs to the machine.

Jen-Hsun Huang showed off a development board on stage with a Pascal GPU in the middle, calling it “a supercomputer the size of two credit cards.”

It’s a tantalizing glimpse into the future of Nvidia GPUs, but regardless of the amount of bandwidth and research into the world-changing possibilities that this cheap supercomputer can offer, I just want to know how smooth it renders Unreal Engine 4 to 4K.

There was noticeably little talk of 20nm Maxwell GPUs at the conference. We’ve seen mainstream releases of the 28nm GM107 GPU in the GTX 750 and GTX 750 Ti, but what about the top-end Maxwell? The dual GPU GTX Titan Z still runs the Kepler based GK110 GPU, is there a high performance Maxwell card at work?

Check more articles in our categories Gaming & News & Anime.

Thanks for visiting we hope our article Beyond Maxwell: Nvidia Announces Their Next-Generation Pascal GPUs

, think about share the article on Facebook, twitter and whatsapp with the hashtag ☑️ #Maxwell #Nvidia #Announces #NextGeneration #Pascal #GPUs ☑️!

Wilbert Wood
Games, music, TV shows, movies and everything else.