Tapping Into the Power of GPU in Mathematica
Last week we posted an item about Wolfram Research‘s partnership with NVIDIA to integrate GPU programming into Mathematica. With NVIDIA’s GPU Technology Conference 2010 starting today, we thought we would share a little more for those who won’t be at the show to see us (booth #31, for those who are attending).
Mathematica‘s GPU programming integration is not just about performance. Yes, of course, with GPU power you get some of your answers several times faster than before—but that is only half the story.
The heart of the integration is the full automation of the GPU function developing process. With proper hardware, you can write, compile, test, and run your code in a single transparent step. There is no need to worry about details, such as memory allocation or library binding. Mathematica handles it elegantly and gracefully for you. As a developer, you will be able to focus on developing and optimizing your algorithms, and nothing else.
Here are a couple of examples to give you a taste of the upcoming feature.
We have often been asked whether it is possible to do 3D volumetric rendering with Mathematica. It’s a natural question, considering that Mathematica has supported DICOM format for a while. There have been a few solutions, but nothing was perfect nor fully integrated with Mathematica—until now. With the power of GPU, you can embed interactive volumetric images into your notebook.
Another example is ray tracing. The following image shows a ray-traced Quaternion Julia set using GPU. We can perform real-time ray tracing and maintain an interactive frame rate on consumer-level graphics hardware.
Of course, this is just the tip of the iceberg. We chose these examples simply because they look beautiful. You will be able to see more examples and applications during the NVIDIA conference, so again, stop by booth #31 if you get a chance.
Not attending the NVIDIA conference, but interested in learning more? There’s still time to register to attend the Wolfram Technology Conference 2010 taking place in October and learn how you can tap into the power of GPU in Mathematica.