Skip to main content

Nvidia new AI brain has eight Pascal GPUs, 7TB of solid state memory, and needs 3,200 watts

At the Nvidia GPU Technology Conference Keynote, Nvidia co-founder and CEO Jen-Hsung Huang revealed the new Tesla P100, an AI-focused chip with some impressive statistics, and the DGX-1, a home for eight of the new chips to live in.

Tesla P100

The Tesla P100 is a Pascal-based chip that’s by far the most ambitious Nvidia, or anyone, has ever undertaken. It packs 150 billion transistors into a 16 nanometer FinFET chip, resulting in an impressive 5.3 teraflops of performance. It also reaches new heights of memory bandwidth thanks to its use of High Bandwidth Memory 2, and the P100 is the first to feature the tech.

Related Videos

Nvidia is completely committed to artificial intelligence and deep learning.

Before you start wondering how to stop your CPU from being a bottleneck, it’s important to remember these chips aren’t built for gaming. They’re very much an enterprise solution built specifically for massive deep learning networks, and they’ll need a home inside a fitting supercomputer.

Importantly, Nvidia has improved the chip’s ability to communicate with other GPUs. This new connection is called NVLink, and it allows far more bandwidth than even PCIe can provide. Nvidia has taken full advantage of that connectivity to produce a new product, the DGX-1.

DGX-1 supercomputer

For users who are looking for something a little beefier than just a single P100, the rack-mounted DGX-1 is powered by a set of eight Tesla P100s. It’s specifically engineered for AI and deep learning development, making it first of its kind. Nvidia claims its memory has as much throughput as 250 x86 servers, and it also has 7TB of SSD storage right on board.


The chips are arranged in a hybrid cube array, allowing them to communicate with each other independently and off-put tasks in a more coordinated way. It’s an almost too simple solution for building an AI network in your enterprise, and it’s capable of handling almost any task thrown at it in short order.

The DGX-1 is already available for pre-order, but start saving up, because the impressive rig will cost you $129,000 when it ships in the third quarter. Nvidia has chosen a small group of hospitals and universities to start out as the early adopters for the DGX-1, and those machines should be rolling out to them soon.

What’s the point?

At the end of the day, eight Pascal GPUs are 12 times faster than the four Maxwell GPU system unveiled at GTC in 2015. That’s a massive increase in speed and power.

It might not be as exciting as a new gaming chip, but it has massive implications on everything from cloud networks, to social media, to self-driving cards and autonomous robots. It’s not for topping 3DMark leaderboards (thought it’d probably do well in Fire Strike!), but that means the all-business Tesla P100 has time to focus on improving technology in other areas.

Nvidia is completely committed to artificial intelligence and deep learning, and that’s only become clearer as more intense high-end products start to appear. That doesn’t mean gaming will hurt, by the way, but we’re clearly approaching the cap for what we need in terms of graphical performance. Instead, the GPUs of tomorrow will be built for AI and deep learning, and gaming will just be a hobby.

Editors' Recommendations

Nvidia’s Pascal will take video memory to a whole new level
Mercedes Benz Concept IAA Project Dash dash

Graphics cards have always required extremely quick memory. That's why for years they've sported GDDR5, while main system RAM has used DDR3. That ends with Nvidia's next generation of graphics cards, currently codenamed Pascal, which will take things to a whole new speed level thanks to the second generation of High Bandwidth Memory (HBM).

We first saw HBM in AMD's Fiji XT-based Fury line of GPUs, and while its bandwidth was impressive, it seemed constrained by the number of gigabytes that could be packed on to the PCB. Pascal will fix that: Some cards in the new line-up sport as much as 16GB of the high-speed memory.

Read more
Madman cuts open Surface 3 Pro to install new 1TB solid state drive
madman cuts open surface 3 pro to install new 1tb solid state drive dumarjr

We've heard of customized computing before, but this is getting a little out of hand.

Back in February, the world caught wind of a new method to upgrade the solid-state hard drive included in the Microsoft Surface 3 Pro that allowed users to push beyond the company's self-imposed limit of 512GB. The technique, pioneered by Mexico City's Jorge Malagon, doesn't involve going through official channels on Microsoft's side. Instead he used a drill to crack open the tablet/laptop hybrid by cutting straight through its metal body.

Read more
The Framework Laptop 16 is officially my most anticipated laptop
framework laptop 16 most anticipated 01

Framework has announced a new addition to its lineup of modular, repairable laptops -- the Framework Laptop 16. It's far from just a larger version of the Framework Laptop 13, though. It takes the idea of modularity to the next level, adding a number of new expansion systems that can be customized to your heart's content -- especially for those who need more powerful graphics.

Before we get to those neat expansion systems, though, just take a look at this design. The two-tone design takes the best of an Alienware gaming laptop and a MacBook Pro, while still feeling totally fresh. I dig it.

Read more