A Closer Look

With its all-black design, ASUS' heatsink looks ultra-stylisch. Five heatpipes which make direct contact with the GPU's surface can be found on the base.


Taking a look at the card without its primary heatsink attached, you can see a secondary heatsink that cools the voltage regulation circuitry. Near the top-edge is a metal bar that is there as reinforcement to make sure the card doesn't sag once it has been installed into a case.


Power delivery requires two 8-pin PCI-Express power connectors. This configuration is specified for up to 375 W power draw.


Near the SLI connector is a switch that toggles between the normal and LN2 BIOS. This switch is only enabled after bridging the "Enable LN2" solder points.


ASUS has included a bunch of enthusiast overclocking options on their Matrix. Here's a list from left to right:

• Molex power connector for the memory defroster (not needed for normal operation). When using liquid nitrogen, the back of the card will freeze with ice due to condensation, which can short circuits and result in cold bugs. The memory heater is activated by connecting a solder bridge and wiring a button to the corresponding XJ16 pins.
• You will find a bunch of unlabeled solder vias near XJ16. These are used to connect the VGA Hotwire feature on a supporting motherboard, but you could also emulate that behavior with trimpots and a DMM.
• The big "safe mode" button will restore the default clocks and voltages if you saved them to the VBIOS using ASUS GPUTweak.
• On the other side of that button is the "enable LN2" solder pad, which needs to be connected to enable the BIOS switch for you to run the second BIOS that is optimized for liquid-nitrogen usage.
• There are six solder vias all the way to the right and near the power connectors of which I'm not sure what they do (I've asked ASUS).

ASUS is using their own voltage controller, which seems to be a rebranded CHiL 8318.


The GDDR5 memory chips are made by Hynix and carry the model number H5GQ4H24MFR-R2C. They are specified to run at 1750 MHz (7000 MHz GDDR5 effective).


NVIDIA's GM200 graphics processor is the company's flagship GPU. It is produced on a 28 nm process at TSMC, Taiwan, with a transistor count of 8.0 billion and a die size of 601 mm².