Nvidia's CUDA Celebrates 5th Anniversary

CommanderFrank

CommanderFrank

Cat Can't Scratch It
Joined
May 9, 2000
Messages
75,399
Five years ago, NVidia introduced the CUDA platform for developers. The platform provides developers with access to the virtual instruction set and memory of the computational elements in CUDA GPUs. The CUDA toolkit has been very successful, being downloaded over a million times and used at universities around the world.


CUDA, which is still positioned against open high-level platforms, especially OpenCL, survived a looming battle with Intel's canceled Larrabee graphics card and floating point accelerator.
Click to expand...
 
I finally found a useful program that actually utilizes CUDA too, after only 5 years :p

http://mirillis.com/en/products/action.html It's like FRAPS but way faster. I drop 20-25FPS in BF3 with FRAPS and 5FPS with Action! Seems like a really nice program. It only uses CUDA for the encoding part though.
 
A Japanese researcher managed to build one of the world cheapest supercomputer thanks to CUDA
http://www.youtube.com/watch?v=hSVo4ejZ7rc

He basically used the commercial nVidia's graphic cards to build his super computer. The setup may not look professional but it does the job.
 
Well, I guess I better play through Alice: Madness Returns to experience the CUDA before Batman is up for preload.
 
And Cuda in 3d studio max and maya fuctions because autodesk is to fucking lazy to support tons of GPUS like games do , so they got in bed with NVida to sell a bunch of overpriced CUDA workstation GPUS that would work just fine on any fucking cheper game class Nvida GPU but they want to be bastards and lazy at the same fucking time. Oh and they didn't write ATI or fireGL code at all



CUDA-card.jpg
 
<iframe width="420" height="315" src="http://www.youtube.com/embed/4zVAR_l2w8k" frameborder="0" allowfullscreen></iframe>
 
I love how "Cube" seems to have some sort of personal investment in this. Like CUDA murdered his grandmother or something.
 
Entertainer said:
Well, I guess I better play through Alice: Madness Returns to experience the CUDA before Batman is up for preload.
Click to expand...

One of my favorite games of this year. It's the only game i've seen where PhysX made a signficant impact to the visuals.
 
fattypants said:
You mean "is"?

Also, Stream is very efficient when people actually decide to use it. In programs that are programmed for both CUDA and Stream, Stream usually wins.
Click to expand...

This is simply bogus information, and I'm calling you out. What you're referring to has nothing to do with CUDA vs. Stream. The cases where AMD running Stream is faster than NVIDIA running CUDA, the reason is most likely that the developer only bothered with a Stream version of their application because they knew their particular math workload mapped well to AMDs architecture. Why would you develop on a less supported platform otherwise? What I'm saying is that by nature of the fact that someone made a Stream implementation at all, it's probably because it maps to AMDs hardware better than NVIDIAs (they have different strengths).

Performance differences are due to how a given problem or workload maps to each company's architecture... not the APIs used to do the work. You can find cases that excel on either company's hardware.
 
Hornet said:
A Japanese researcher managed to build one of the world cheapest supercomputer thanks to CUDA
http://www.youtube.com/watch?v=hSVo4ejZ7rc

He basically used the commercial nVidia's graphic cards to build his super computer. The setup may not look professional but it does the job.
Click to expand...

Why are most supercomputers not built using consumer level parts? I mean, what is the difference with Quadro graphics cards that makes them so expensive?
 
You must log in or register to reply here.
Back
Top