   |
||||||||||||||||||||||||||
|
||||||||||||||||||||||||||
|
:
News
: Archives
|
|
The GeForce 7600 GT |
|
|||||||||||||||||||||||
|
eVGA Superclock Style |
||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
|
by Josh Walrath |
||||||||||||||||||||||||||
|
Anti-Aliasing performance I found a very interesting case where enabling anti-aliasing does not give a large performance impact. Running a custom Half Life 2 demo at 1024 x 768 with max quality settings did not have an appreciable framerate drop when enabling anti-aliasing.
I was quite shocked that HL2 remained CPU bound throughout the different AA settings. Losing 2 fps when going from no AA to 4X AA is pretty astounding. 8XS is eminently playable at this resolution! I also wanted to see what kind of effect transparency AA had on performance at the same quality level.
Again we don’t see much of a performance drop in this application when going from no AA to full 4X TSAA. Now, this does not mean that every application out there will exhibit the same behavior, as it is obvious that HL2 at this quality level is CPU bound. Still, I thought this was a very interesting case considering how popular HL2 is and how common the 1024 x 768 resolution is.
The 7600 GT (top) is significantly smaller and simpler than the 6800 GT (bottom) though the 7600 GT consistently outperforms the older card. Anisotropic Filtering Performance I decided to take things up a notch with this test. I used the same HL2 demo as the AA benchmark, but I increased the resolution to 1600 x 1200 with 4X AA enabled.
There is a very appreciable drop in performance when enabling 16X AF. It is not crippling, but it is significant. Like 30% significant. PureVideo Performance While no changes have been made in the hardware, NVIDIA has been working on the quality of the drivers when it comes to video performance. The latest beta Forceware drivers (and the 84.19 drivers that will be available at launch) add H.264 acceleration support as well as high definition MPEG-2 inverse telecine and spatial temporal de-interlacing. I broke out the HQV 1.4 benchmark and compared the results to an older driver and a 6600 GT. The 6600 GT with the 82.98 drivers scored a 53 on the test, while the 7600 GT with the 84.17 drivers scored slightly higher with 63. NVIDIA has done some improvements to PureVideo in playback with this latest driver. When playing back H.264 content, CPU usage went from 100% on the 6600 GT and older drivers to around 60% on the 7600 GT and 84.17’s. Quality looked very good and I had no complaints. The only real issues I ran into were the lack of noise reduction in some videos as well as non-standard cadences. NVIDIA has excellent support for the 3:2 and 3:2:3:2:2 cadences, but the other cadences suffered greatly. For standard DVD movies filmed in live action, the replay quality was superb. Animation and anime were a different story though. The PureVideo decoder did not handle those non-standard cadences very well, and I noticed some nasty temporal issues with anime. Still, the overall video quality was excellent, and it would be hard to go wrong with this product in a home theater PC. Overclocking Since the 7600 GT CO Superclocked is already overclocked out the box, I had little success getting much more out of the card. I was able to take the core from 600 MHz up to 619 MHz. Anything over that would cause lock ups. The memory was able to go from 780 MHz up to an astounding 791 MHz. 3D Mark 2006 went from 3489 up to 3612. This isn’t a huge jump, and this usually came out to be about a half a frame faster on each scene than the standard 600/780 setting. Users expecting to hit 650 MHz or above with this particular product will probably be disappointed.
If you have found this article interesting or a great help, please donate to this site.
Copyright 1999-2006 PenStar Systems, LLC. |
||||||||||||||||||||||||||
