Nvidia GTC 2015 Live Blog

[[Coi] Mr.CLuBBeR]

Titan X announced to be $999

Nvidia's GPU technology conference is starting shortly and we're on location in San Jose prepping to do a live blog on the event. Refresh your browser for updates as the conference progresses.
9:11am PST: Video introduction begins: Shows footage of sci-fi movie clips (2001: A Space Odyssey), videos that highlight medical research. Video showcase robots and vehicles that drive humans. The video ends with "the future is here" and Nvidia CEO Jen-Hsun Huang walks out.
9:15: Jen-Hsun Huang says four things will be talked about: a new GPU and deep learning, a "very fast box" and deep learning, roadmap reveal and deep learning, and self-driving cars and deep learning. Deep learning appears to be a theme here...
9:17: Jen-Hsun Huang recaps the last year in visual computing. Says gaming is big. Brings up Nvidia Shield console announcement from GDC.
9:18: The days of dials, knobs and buttons are gone for cars, says Jen-Hsun Huang. It will all be digital, he says. Cars will be smart, he suggests.
9:22: Huang says since 2008, there has been a 10X growth in GPU computing.
9:27: Huang formally announces Titan X GPU. Video plays introducing us to the chassis of the card. Specs: 8 billion transistors, 3072 CUDA cores, 7 TFLOPS SP/.2 TFLOPS DP, 12GB Memory.
9:31: Huang showcases what Titan X can do by showing Epic's new Unreal demo which is a forest area that has been rendered out 100 square miles (larger than Silicon Valley, Huang says). We've seen this video at GDC before, but it never fails to impress. A boy and his kite fly through a gorgeous virtual world. It looks like a CGI film (except its supposedly being rendered real time on a Titan X).
9:36: Huang talks about Titan X for deep learning. Says its faster than a 16 core Xeon CPU, which took nearly 43 days) under the training AlexNet benchmark. Titan X, on the other hand, took only 2.5 days with a middleware called cuDNN.
9:40: Huang announces Titan X will cost $999
9:46: Huang talks about how machine learning has been used in the late 90s to allow machines to detect human writing, which we still use in banks and the like today.
9:47: Huang recaps that on February 6th, 2015, Microsoft used a computer that could recognize images better than a human.
10:06: Huang puts on a slide showcasing dozens of tech companies using GPU accelerated deep learning. These companies include Facebook, IBM, Microsoft and more.
10:07: Huang explains that deep learning is being used in medical research. A couple of examples include: predicting the toxicity of new drugs and understanding gene mutation to prevent disease.
10:10: Huang says we're just at the beginning of deep learning. Showcases research done at Stanford university which is able to use a computer to create a sentence which descibes an image. The sentence reads, "a bird perched on a branch of a tree," as we see an image of a bird perched on a branch of a tree. That's pretty impressive.



More...