A Closer Look

The cover of the GeForce GTX 295 is just a simple hole metal cover which serves no purpose other than to direct the airflow and protect the card. The central cooling assembly is mostly metal and seems to be a complex piece of engineering. It cools both GPUs and memory chips on both sides at the same time. Also cooled are minor components like voltage regulator ICs.


Two GTX 295 cards may be combined for Quad-SLI to give you even more performance.


The card has a six-pin and one eight-pin power connector. Both are required for operation.
When I looked at the card for the first time, I saw the sticker and the single 6-pin connector above it and wondered "Oh? Only one six pin power connector required?".


The GDDR3 memory chips are made by Hynix and carry the model number H5RS5223CFR-N0C. With a latency of 1.0 ns, they are specced to run at 1000 MHz.


NVIDIA is using their own NF200 PCI-Express bridge chip to interconnect both GPUs. AMD uses a chip from another manufacturer for their HD 4870 X2. NVIDIA's chip offers some additional features which are supposed to help with SLI performance, we have also seen several motherboard designs on which the NF200 is used.


NVIDIA has separated the display output logic from the GPU on the latest chips. So in order to drive the two DVI outputs a single NVIO chip is required. Since the card also has one HDMI output on the other PCB, another NVIO is needed there as well.


Here you can see the two GT200b GPUs that power the GTX 295. Each which is made in a 55 nm process with 1.4 billion transistors.