First Impressions and Overclocking
Take a Look
![NVIDIA GeForce 256 DDR Hands-On Preview [ GeForce 256 DDR @ 640 x 480 ] > View Full-Size in another window.](images/ddrboard-s.gif) GeForce 256 DDR
|
|
![NVIDIA GeForce 256 DDR Hands-On Preview [ The back @ 640 x 480 ] > View Full-Size in another window.](images/ddrboardback-s.gif) The back
|
|
As you can see from the pictures, the DDR card has Infineon HYB39D32322TQ DDR SGRAM. Our other GeForce SDRAM reference board has the same 5ns ESMT M12L16161A-5T SDRAM that we've seen on a couple TNT2 Ultra boards. The DDR card has an s-video out with a Brooktree chip and what appears to be a digital flat panel output.
![NVIDIA GeForce 256 DDR Hands-On Preview [ Infineon DDR SGRAM @ 219 x 155 ] > View Full-Size in another window.](images/ddr-s.gif) Infineon DDR SGRAM
|
Overclocking
Overclocking the reference GeForce 256 DDR board was a bit difficult. We had to download the newest version of Entech's Powerstrip program which has preliminary GeForce support. Unfortunately, we discovered that Powerstrip doesn't have support for the DDR version yet. We were able to overclock the engine clock speed, but were unable to overclock the memory. Even after mucking around in the pstrip.cfg file, we were unable to get Powerstrip to accept any new memory settings.
Oh well, at least we can still overclock the core. After a few overclocking attempts we found that the chip was stable at a nice 130MHz. The overclocking the engine didn't really help the Quake3 timedemos very much, giving less than 3% at the high resolutions, and nothing at the low resolutions. Obviously, memory bandwidth was the bottleneck at higher resolutions, but the lower resolution bottleneck was unknown. We'll talk a little more about our low resolution Quake 3 experiments in the benchmarks.