A TITAN for a Titan: NVIDIA CEO Jen-Hsun Huang Presents New TITAN X to Baidu’s Andrew Ng

by Brian Caulfield

Everyone has heroes. Baidu Chief Scientist Andrew Ng, one of the pioneers of deep learning, is one of ours.

So NVIDIA CEO Jen-Hsun Huang chose to light up a meetup of deep learning experts on a glorious mid-summer evening at Stanford University to unveil NVIDIA TITAN X, our latest top-end GPU, by presenting it to Ng.

The audience of more than 500 academics, researchers, and students — gathered in an open, airy hall at Stanford’s faculty club — quickly dropped their canapes and picked up their smartphones to snap pictures of the moment.

“When I get excited, crazy stuff happens,” Jen-Hsun, clad in one of his trademark leather jackets, told the audience. “We wanted to bring supercomputing power into the GeForce channel, so everyone in academia can benefit.”

A Hero to Us All

Jen-Hsun then presented the first TITAN X to Ng, reading an inscription on the TITAN X describing Ng as “a pioneer, an amazing scientist, and a hero to us all.”

GPUs — along with the torrents of data unleashed by the Internet — have played a key role in the deep learning boom led by researchers like Ng that is shaking the world to its foundations.


Back in 2012, Ng helped jumpstart the artificial intelligence field by using GPUs to help build a deep network of artificial neurons — then running 10 million YouTube videos through the system to train one of the first deep learning systems. Since then, the speed of deep learning systems has increased fifty times.

The results of breakthroughs made by researchers such as Ng are set to upend entire industries. So it’s fitting that Ng, who now leads Baidu’s efforts to put deep learning to work for everything from voice recognition to image search, will be among the first to receive a TITAN X.

A Transformative Moment

“Just as electricity 100 years ago transformed industry after industry after industry, I think AI powered by deep learning will now do the same,” said Ng , who is also an associate professor at Stanford, speaking to the standing room only crowd. “It’s hard to think of an industry that will not be transformed by AI in the next decade. ”

It’s a field where access to cutting-edge infrastructure is critical. “If you’re a machine learning researcher having access to a machine that is 2x as fast means that you are 2x as productive as a researcher,” Ng said.

Rarefied company indeed, one in which the TITAN X fits right in. The TITAN X is the ultimate graphics card. Whatever you’re doing, this groundbreaking NVIDIA Pascal-powered GPU gives you the power to accomplish things you never thought possible.

“One of the things I admire about NVIDIA is it’s breaking new ground, and not just chasing profits,” Ng said.

We packed the most raw horsepower we possibly could into this GPU. Driven by 3,584 NVIDIA CUDA cores running at 1.5GHz, TITAN X packs 11 TFLOPs of brute force. Plus it’s armed with 12 GB of GDDR5X memory—one of the fastest memory technologies in the world.

Researchers Gasp

Audience members at the gathering gasped and hooted as Jen-Hsun detailed the TITAN X’s capabilities. The excitement only built as Jen-Hsun read out seat numbers to give away TITAN Xs to people in the crowd.

“This is the first product launch I’ve ever done where I’m standing next to guys in t-shirts,” Jen-Hsun said.

“This is insane, I’m actually speechless, I didn’t expect something like that,” one of the event’s organizers stammered after the TITAN X was unveiled. “This is one of the craziest product launches ever in the history of technology.”

Similar Stories

  • Artificial Intelligence Meetup

    Thank you Jen-Hsun and Andrew Ng for joining us this evening @SFAImeetup. It was an evening full of surprises #TITANX

  • krazyfrog

    Whoever wrote this needs to calm down a bit.

  • Jason Cordova

    Glad ive been with nvidia for years since i snapped in my very 1st gpu trying to upgrade and learn all about this stuff. Looks like its only gonna get better from here.

  • Sven Meyer

    So, what about FP16 performance which is quite useful for neural nets? Has it been crippled like on the GP104 (GTX1080) resulting in 1/64 of FP32 performance?

  • Charlie

    I think you need to GET MORE EXCITED!!!

  • Bob Austin

    no hbm2? 🙁

  • http://daow.net DAOWAce

    No HBM/2 and only 12GB of VRAM. That is disappointing.

  • Harawanagangsta

    Titan XP please.

  • Habib Arshad

    What lol?
    Pascal makes fp16 x2 of fp32
    Im sure your talking about fp64?
    I think that it is enable though partially….

  • Sven Meyer

    I am talking about FP16! Only the expensive (>6000$) GP100 Pascal GPU does FP16 twice as fast as FP32.
    There are different versions of Pascal chips, with different performance characteristics.
    Pascal GP104 (GTX 1080) has on 1/64 of FP16 compute units than for FP32, hence only 1/64 of performance.
    Now the question: Now many FP16 compute units does Pascal GP102 (NVIDIA Titan X) have?

  • Sven Meyer

    There is a shortage in available HBM2 chips, so you will not see them any time soon in consumer cards. The HBM2 chips all end up with the 6000$+ GP100 Pascal GPUs at the moment … maybe next year (with Volta).

  • Habib Arshad

    That doesnt matter cause you still get 480gb/s where hbm2 is near 700gb
    All other that matters is size and power which imo is not a big case

  • http://daow.net DAOWAce

    What doesn’t matter? The VRAM amount or it not being HBM?

    If VRAM: Yes, it damn well matters. We’ve already had some games use near 8GB VRAM, and higher resolution games need more VRAM regardless. Forget about modding things with 4K or 8K textures; you have to cut back because GPUs just haven’t had enough VRAM (NVIDIA at least; AMD GPUs have had 8GB while NVIDIA was stuck on 4GB).

    And for HBM: It matters too; less memory bandwidth is just worse overall for every application. Memory bandwidth bottlenecks (across the entire system) are not easy to diagnose and can cause performance issues that make no sense.

    It was rumored the Ti or Titan card would have 16GB+ of HBM2 memory. It doesn’t; so we’re all disappointed. Not even HBM1 either.

    Some people are saying this isn’t the full Pascal chip, that the Ti and “NVIDIA Titan X” v2 will be using the full chip.. and if so, they’re not coming for a while (at least, I wouldn’t expect them).