Sapphire X1600 Ultimate vs MSI NX7600

System Setup

To test out the capabilities of these cards, we used one of our Intel reference rigs. This consists of:

Intel Pentium Extreme Edition 965 (3.73GHz, 2x2MB L2 cache, 1066MHz FSB); ASUS P5WD2-E Premium (Intel 975X/ICH7R chipset); 2x512MB Corsair XMS2-8500 DDR2 memory (running at 800MHz with 4.0-4-4-12 timings); Maxtor DiamondMax 10 250GB SATA Hard Disk Drive; Windows XP Professional Service Pack 2;

Video testing

There are a couple of steps you need to do to get video acceleration working properly on both platforms.

For NVIDIA, you need to go and grab the latest ForceWare - we used official driver 84.21, which contains the PureVideo codec. If you're running Windows Media Center, make sure you download the right version of the driver.

You then need to get an officially supported MPEG 2 decoder and an officially supported H.264 decoder if you plan on watching high definition.

For ATI, you need to get the latest version of the Catalyst Control Panel and driver - we used official driver 6.5, which has Avivo built in. Like NVIDIA, you then need to go and get supported MPEG 2 and H.264 decoders.

Both NVIDIA and ATI recommend Cyberlink PowerDVD for decoding duties. The basic PowerDVD 6 which many of you will have found packaged with your last graphics card purchase contains an MPEG 2 codec that can be accelerated by both. For an extra $25, you can buy a downloadable add-on to support H.264 decoding on the GPU. This codec can then be used either in PowerDVD itself, or by Windows Media Player. Alternatively, the new version of PowerDVD, version 7, comes with that support built in and costs $70.

The one thing to note is that NVIDIA doesn't expose any tweakability to PureVideo in its latest official drivers. ATI, by contrast, provides some cool options to pick de-interlacing standards and tweak the output picture to your preference.

HQV

This is an industry-standard benchmark designed to test the optimisation of MPEG 2 playback. MPEG 2 is the codec used on current DVDs, as well as on digital TV broadcasts.


Standard playback with no acceleration can leave edges looking aliased and motion blurry. Good acceleration should smooth out edges and leave video looking smooth and sharp.

There are a number of video tests in HQV, designed to detect things like anti-aliasing, correction of frame rates and colour reproduction. It's essentially the 3DMark of video benchmarks - great for giving you numbers, but of limited real-world use.

We put both cards through the HQV test and ended up with the same score for image quality. Both sides claim that their cards should win the shootout on HQV score, but we were unable to discern a noticeable difference on our 21" CRT monitor at 2048x1536. There was also no discernible difference when we hooked up to a 1920x1080 42" LCD panel. Both cards did a thoroughly good job of smoothing out the image quality of the reference DVD stream. Neither, however, does any image sharpening.

It's worth noting that whilst both teams make a big deal out of 'cadence' detection and 'pulldown' detection, these features are only relevant when watching Region 1 DVDs encoded in NTSC - there's no effect on PAL (Region 2 DVD, or Freeview) footage, which doesn't need the correction.