ATI Radeon HD 5970 review

What's the point..really?

Only the most hardcore of gamers are likely to be in the target demographic for a card like this. For not much more you could get an XBox & an HD TV. I know what I'd buy, and it isn't this.

By mikeeJ on 19 Nov 2009

No business Apps for high end graphics?

Are PC games the only application for graphics cards, are there no business where the higher performance could equate to productivity gain?

By milliganp on 19 Nov 2009

Re: No business Apps for high end graphics?

@milliganp: Computer aided design (CAD) work requires high end graphics cards, but they tend to use cards specifically designed for that purpose. You can also do number-crunching (science/statistics) stuff, but there you need a general purpose graphics card (GPGPU). Most business apps work in 2D, so until Microsoft releases Word3D, or Second Life takes off for businesses, a gaming card isn't necessary at work.

By rupert_giles on 20 Nov 2009

Most Powerful and Advanced HD5970

That is too great, i got a video card of HD5870,from: http://www.it-battery.com/his-hd-5870-1gb-256bit-g
ddr5-pcie-graphics-card-p-1916.html

it is too great,raises this sword for gamers,that is a great felling, now HD5970 comes into market, much powerful than HD5870.

By Roynorton on 20 Nov 2009

Re: No business Apps for high end graphics?

@milliganp & @rupert_giles

As far as I'm aware GPGPU software is mainly used for engineering, physics, chemistry and mathematics - I would LOVE it to be available for statistical analysis, given that some procedures are simply too slow to run (or crash) when using a CPU.

By Peter_Tennant on 22 Nov 2009

Very much the point...

We should applaud innovation and welcome the fact that this card exceeds anything current games demand. Remember when games exceed cards? Isn't it better when cards exceed games, that way the price of cards will tend to come down and we'll all be more likely to play the newer games at higher settings.

By simontompkins on 22 Nov 2009

GPGPU Usage

The Stream API works for this GPU, so there's really not much distinguishing it from GPGPU except perhaps poor marketing. As for statistical software, the only reason science apps can use GPUs like Nvidia is because someone took initiative to code for it using an appropriate API. Nothing is keeping it from being useful for stats algorithms except algorithms than cannot be parallelized (usually not the case).

I personally plan to buy a Crossfire dual 5970 and use it as a GPGPU with Stream for my fluid dynamics applications. 9 Teraflops (or even 20% efficiency to get 2 Teraflops) is just beyond comprehension for a single machine maxing out at only $4,000. Needless to say, my mouth is indeed watering.

Given recent progress, I don't think it will be long before High Performance Computing steals the show because you'll never make a processor that's faster for HPC applications demand.

By mrnorman on 26 Apr 2010