Zotac 9800GTX+ AMP!
Author: Vedran Dakic
Date: 07 Nov 2008

When ATI released their 4800 series, or 4850 card to be more precise, NVIDIA was forced to react. The 4850 outperformed their every offering except the new, expensive GTX 200 series cards, so their first move was to slash the price of the 9800GTX. The card we have here today was the next step. It is built around a die shrunk G92 GPU, and named 9800GTX+. Essentially, it is an overclocked 9800GTX. Zotac’s AMP! Version is an overclocked version of an overclocked card. Let’s see how it did in our benchmarks.

The GPU
There is not much to be said about the G92 GPU. It has been powering NVIDIA graphics cards since the 8800GT. It has 128 shader processors, 16 ROP units and a 256 bit memory bus. Its first incarnation was built by TSMC using a 65 nm production process, while this new version uses a 55 nm process. It is only a die-shrink, and all of the features are there. Smaller production process enables two things. The first one is lower price, since you can fit more chips onto a 300 mm silicon wafer. The second is lower heat dissipation, which enables higher clocks, quieter fans and more efficient power consumption.

The G92 in a stock 9800GTX+ card is clocked at 738 MHz for the core and 1836 for the shaders. Zotac upped that to 756 and 1890 MHz. The 0.8 ns memory from Samsung operates at 1150 MHz, or 2300 MHz effective because of the double data rate of the 512 MB GDDR3.

The card
There is no difference between the normal and + versions of the 9800GTX. The cooler is the same, as well as the PCB. The only thing you notice is the smaller surface of the G92 die on the second card. The memory chips are different, since these operate at higher frequencies. Zotac’s card is not built on a custom PCB, and without a sticker, it could just as well be an NVIDIA reference card.

Here's a picture of the box and the card: