Introduction

Like CPUs, graphics processors have seen remarkable growth in their performance over the last few generations. In fact, they have grown so fast that the hardest job in the industry is not in finding new ways to improve their performance further but to find new tasks to use up all that massive processing power! Just take a look at how far we have come in 3D graphics!

In a space of just four years, NVIDIA's top graphics card models have doubled in pixel fill rate and more than tripled in texture fill rate and memory bandwidth. With so much processing power, the latest graphics cards from NVIDIA or ATI have no problem handling super-high resolutions and even "expensive" 3D features like anti-aliasing. As a result, both NVIDIA and ATI have encouraged developers to make use of their graphics processors for general purpose computing.

This has resulted in the development of codecs and tools that allow the encoding and decoding of high resolution videos using the graphics processor. Even distributed computing projects have jumped on the bandwagon, developing clients that make use of the graphics processor's tremendous processing capability. That's not the limit of what the GPU can do though. NVIDIA recently revealed their OptiX ray tracing engine, which allows you to harness the tremendous processing power of NVIDIA's Quadro graphics cards to render complex 3D scenes.

Both NVIDIA and ATI graphics processors are based on a unified architecture with programmable stream processors that can handle not only traditional graphics (vertex, geometry, texture) processing but also general purpose computing operations. However, NVIDIA have an ace up their sleeves that currently gives them an edge over ATI and that ace is called PhysX.

What Is PhysX?

PhysX refers to both the proprietary realtime physics engine, as well as the Physics Processing Unit (PPU) developed by AGEIA, a fabless semiconductor company that specialized in the acceleration of physics calculations for use in video games. Founded back in 2002, it was acquired by NVIDIA in February 2008, a fact we actually predicted way back in 2006! Even back then, we could see that it made sense for NVIDIA (or ATI) to buy over AGEIA and their PhysX technology.

Buying AGEIA allowed NVIDIA to integrate PhysX into CUDA (Compute Unified Decide Architecture), the parallel computing architecture used in their GPUs. In one swoop, all NVIDIA graphics processors from the GeForce 8 series onwards were instantly able to accelerate physics calculations for games and applications that supported the PhysX engine.

Instead of using a dedicated PhysX PPU, NVIDIA uses some of the stream processors in the GPU to process the physics calculations. This allows NVIDIA to divert some of their GPUs' tremendous processing power to physics calculations without using a separate PPU. The result? Physics acceleration by the GPU at no additional cost.

Of course, PhysX is NVIDIA's proprietary technology. Only NVIDIA graphics cards are capable of accelerating games and applications that support PhysX. This gives them an important leverage over their competitor, ATI, who has made a remarkable comeback on the back of their Radeon HD 4000 series of graphics cards. As popular as the current ATI graphics cards, they cannot accelerate games and applications that support PhysX.

But just how important is PhysX acceleration? That's what we are about to find out!

Support Tech ARP!

If you like our work, you can help support out work by visiting our sponsors, participate in the Tech ARP Forums, or even donate to our fund. Any help you can render is greatly appreciated!