A Closer Look

EVGA uses exactly the same fan as the original NVIDIA GeForce 8800 GTX. The air is sucked in and routed away in a 90° angle. This blower fan approach works very well and keeps the card cool and noise levels down. EVGA added a little plastic label that says "EVGA e-GeForce 8800GTX Series | ACS³". This label is protected from scratches by a transparent plastic film, very nice attention to detail here.


Two SLI connectors are available for future configurations like three or four cards in SLI. At this time you can only do SLI with two cards using either of the SLI connectors.


A big metal plate on the back of the card adds extra cooling, because some heat travels from the GPU through to PCB to the other side.


Removing the cooler is several steps. It seems to me that EVGA kept the original cooler base (grey metal, fins and fan) and changed only the black metal part on top of it. The grey patch on the black metal is a thermal pad that sits on top of the heatsink fins and allows additional heat to be dissipated through the cooler's outer shell.


Two power connectors are required for operation of this card, adapters are included.


EVGA did not change the PCB layout itself, so the card is as long as the NVIDIA 8800 GTX. Exact length is 27 cm. Having the power connectors angled 90° helps with the length requirements.


NVIDIA "still" uses GDDR3 memory with 1.1ns cycle time which should be good for about 900 MHz. Having 768 MB of GDDR4 on the card would be excessively expensive for only rather small gains. Not to mention that ATI keeps buying all GDDR4 stock.


When the cooler is removed you can see just how big the GPU is.


The GPU used is the G80 revision A2 which is the latest available.


If you look closely, you can see that the big Grey chunk of metal is not the actual die, but a heatspreader which protects the die from damage during cooler mounting.


On the GeForce 8800 the TMDS display output logic has been moved away from the core into a separate chip, made by NVIDIA. This allows greater design flexibility for future configurations. When several GPUs are working together you need only one set of outputs. If you had a couple of GPUs with their own unused output logic this would mean wasted silicon space and money. Also a physics card does not need any display output logic.