bit-tech.net

Nvidia launches GPGPU investment scheme

Nvidia launches GPGPU investment scheme

Nvidia's GP-GPU investment scheme offers up to $5 million to start-up companies looking to develop general purpose computing solutions on Nvidia hardware.

Nvidia is doing its bit to encourage innovation in the technology sector with the founding of a venture capital fund for start-ups looking to develop GP-GPU technologies.

As reported over on ExtremeTech, Nvidia's GPU Ventures Program aims to provide support – financial and otherwise – to companies looking to create products and services based around general purpose GPU computing – the company's CUDA solution, in other words.

Nvidia has stated that the programme aims to “evaluate companies that leverage the GPU for both consumer and professional applications in all areas such as video and image enhancement, scientific discovery, financial analysis, and 3D interfaces.” Each qualifying start-up will receive between $500,000 and $5 million to invest in development, marketing, and product distribution.

Although the GPU Ventures Program is new, Nvidia's interest in general purpose computing on the GPU isn't: since introducing CUDA back in 2007, the company has invested money in Acceleware, Elemental Technologies, Keyhole Corporation, Mental Images, MotionDSP, and Right Hemisphere – all of whom use Nvidia's hardware to accelerate their software. The new programme simply offers an easier way for start-ups to bring themselves to Nvidia's attention.

Jeff Herbst, Nvidia's vice president of business development, called the programme “a huge opportunity for young ambitious companies basing their businesses around the GPU,” and revealed that the company hopes such start-ups will “fuel the continued growth of the GPU platform.” Pledging to offer “financial, marketing, and other support to help start-up companies realise their full potential” Herbst called for “interested entrepreneurs, venture capitalists, and others to reach out to us with their ideas.

With more and more companies looking to offload processing on to the increasingly powerful GPUs currently on the market, it looks like Nvidia is doing everything it can to ensure that its products are chosen as the de facto GP-GPU standard.

Can you think of an idea for GP-GPU computing that could get you a share of Nvidia's investment capital, or should graphics manufacturers leave the general purpose stuff to the CPU? Share your thoughts over in the forums.

8 Comments

Discuss in the forums Reply
perplekks45 17th March 2009, 14:36 Quote
Someone's trying to push their technology...
Not a bad thing though. Badaboom for example is awesome for video encoding. Let's see what else is coming.
s3v3n 17th March 2009, 15:29 Quote
Quote:
Originally Posted by perplekks45
Someone's trying to push their technology...
Not a bad thing though. Badaboom for example is awesome for video encoding. Let's see what else is coming.

Don't know if they've updated Badaboom but I tried the program about 2 months back and the lowest bitrate was like 500kbps, which to me is just a waste of space. I'm assuming most people are encoding to play the videos on their portable media device with a screen of like 320x240, which only need like 200kbps. Hell, my encodes for my PSP at 480x272 only needs 250kbps and still get a good picture quality.
perplekks45 17th March 2009, 15:35 Quote
Waste of space? Well, maybe but if you re-encode videos in HD there's just nothing that matches the speed of Badaboom.
wuyanxu 17th March 2009, 15:49 Quote
that's good news, at least for nVidia users.

but without opening up their platform, it's never going to take off.
Mentai 17th March 2009, 18:49 Quote
If it does take off do you think AMD will finally give in and pay the licensing fee to run it on their hardware?
Just bought a 4870 1gb :P
Turbotab 17th March 2009, 19:23 Quote
Is this Nvidia's attempt to combat OpenCL? I'm sure the financial incentive is good, but wouldn't it make more sense to develop apps that target the widest possible user base?
Saivert 20th March 2009, 05:40 Quote
OpenCL isn't the only API. You got DirectX 11 Compute Shaders coming as well.

Besides NVIDIA has a good working solution for GPGPU that you can use now.
I don't see anything from OpenCL yet. It's just a spec atm.
Sebbo 20th March 2009, 13:54 Quote
according to the Register, to receive the funding the product doesn't even need to be CUDA based, or run solely on nvidia hardware.
what i find slightly dubious however, is that what you submit to nvidia is basically going to be non-confidential, so unless you have patents etc guarding your ideas, what's to stop them from taking an idea and running with it, especially if they have more resources than you? "Trust" isn't a very good safety line
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums