Talk about a dream assignment. Benedikt, an affable 26-year old intern at NVIDIA, had just been pulled aside by his supervisor, Rochelle, who is a manager in NVIDIA’s 3D vision software team. He was thrilled.
Benedikt — who is working on a doctoral thesis in power efficient gaming at the Technical University of Munich – would be given development boards with early versions of the NVIDIA’s hyper-efficient new GPU, code-named Kepler, in order to build something gamers have long dreamed of – the ability to capture their greatest gaming achievements in real time with no lags, no delays, no stutter.
And he would be able to call on colleagues in Canada, France, and India to help him put it together.
Benedikt would go on to build the first, early version of a feature that – after years of polishing – has woven into NVIDIA’s GeForce Experience service for distribution to more than a million GeForce gamers this summer. Talk about putting theory into practice.
Forget the silly hats and sillier games of Quidditch portrayed in the latest Owen Wilson-Vince Vaughn comedy, “The Internship” about life as an intern at a Silicon Valley tech company.
Interns at NVIDIA have a very different experience, one that surrounds them with passionate, talented colleagues, gives them access technologies too new to have made their way to any classroom, and puts them to work on projects that began long before they join NVIDIA, and continue long after they left.
No make work here.
“We want the interns to be treated as regular employees,” says Monica, who helps run our intern recruiting program. “They work on what we’re doing, there’s nothing they don’t get exposed to.”
Real Work, Real Pay
Benedikt’s tale is typical of a program that aims to put interns in the middle of the action. We hire between 300 and 400 interns every year in the United States – all paid – and plug them into teams working on projects at the very cutting-edge of the fields they’ve studied.
“Before I joined NVIDIA I only had a basic and mainly theoretical understanding of current graphics architectures and how the Windows graphics driver works,” he says. “I had to learn a lot about H.264 encoding, Windows driver development, the windows audio infrastructure and especially about Windows debugging at the driver level.”
That kind of intense, roll-up-your sleeves experience draws 15,000 applications a year for our internship program from top engineering students around the world. Interns meet every week to hear from some of the NVIDIA veterans who helped lead the engineering team that put together our SHIELD portable gaming device; our Chief Scientist; and our CEO, among others.
Makes sense, given that in a few years some of these interns will be helping run the company. The program is an integral part of NVIDIA’s recruiting program: twenty-five percent of our new hires were once interns. And our management team is peppered with former interns.
Where Silicon Meets Software
They’ll be put right in the middle of the action from the start. Benedikt’s project is a perfect example. ShadowPlay’s capabilities are rooted deep in the architecture of NVIDIA’s Kepler-class GPUs, Rochelle explains.
“All the pieces were there,” she says.
Introduced in 2012, Kepler is designed to wring more work out of less power. It’s also built for a more connected world, and the H.264 encoder needed to move rich media content in real time was designed into Kepler’s silicon from the start.
Benedikt completed the first prototype of ShadowPlay in 2011 – before the first Kepler-based GPUs were released – using pre-production boards. Then the same colleagues who collaborated with Benedikt on his project continued working on the project.
By late 2011, the result was already silky smooth. “I didn’t think it would turn out to be so easy to use,” Rochelle says. “The ease of use was just so astounding.”
Ready for Prime Time
So in January 2012, thanks to Santanu, Rochelle’s manager and currently NVIDIA’s Director of Cloud Streaming and 3DVision software, the project was brought to NVIDIA Chief Executive Jen-Hsun Huang for a demo. The name ‘ShadowPlay’ was coined during the demo and the technology was put on the roadmap for our new GeForce Experience service, itself more than six years in the making.
Over time, more software engineers from the audio, video, user experience, and user interface teams at NVIDIA’s Santa Clara headquarters and at NVIDIA’s Pune Design Center (PDC) got ShadowPlay ready for production, building out its functionality, and refining an interface as slick as the silicon behind it.
Available for free to anyone with a GeForce GTX 600 or 700 series GPU, ShadowPlay makes it possible to capture your gaming exploits without slowing down the action, or generating hard-drive hogging video files.
And users can decide how much of their gameplay they want to save, so if they pull off a stunning headshot they can go back and review the action, save it, and share it with their friends in the H.264 video format with just a few clicks.
In short, Benedikt’s project is ready for prime time. Upon its release, ShadowPlay will be available to millions of users. And now, with a new crop of interns on the way, we can’t wait to see what they tackle next. Like every manager at NVIDIA, Rochelle has a long list of projects. None of them involve making copies, or playing Quidditch.