facebook rss twitter

Review: NVIDIA's GPGPU ambition coming to fruition?

by Parm Mann on 2 December 2008, 14:18

Tags: NVIDIA (NASDAQ:NVDA)

Quick Link: HEXUS.net/qaqcq

Add to My Vault: x

Introduction

Gone are the days when a graphics card served the sole purpose of pushing out playable frame rates in all the latest games.

Today, graphics cards from the likes of NVIDIA and AMD need to do more in order to a) warrant some of the high asking prices and b) keep ahead of the game before Intel arrives with Larrabee.

Throughout 2008, NVIDIA has gradually shifted its GPU focus from raw frame rates to the promise of a graphics card that can do a whole lot more. It's called general-purpose computing on graphics processing units (GPGPU), and it means to tell you that your NVIDIA graphics card could be used to carry out computations that are traditionally performed by the CPU.

With modern graphics cards boasting performance in excess of one trillion floating-point operations per second, it makes implicit sense to have the GPU performing other duties when possible. After all, it's something of a waste to have a Ā£300 graphics card that when asked can bring a game to life but sits around idling at all other times.

We've been hearing about the potential of GPGPU for many months, years even, but it wasn't until NVIDIA teamed up with Adobe that we began to see the results. Adobe's Creative Suite 4 became one of the first mainstream software products to take advantage of the parallel processing capability of NVIDIA's GPUs, and with Adobe leading the way, the adoption of GPGPU support is gradually making its way into various other software titles.

NVIDIA hopes the wider use of GPGPU will show the usefulness of NVIDIA PhysX and CUDA technologies, so let's find out if they do.