Design and Features
Nvidia touts its GeForce GTX 690 as the acme of computing technologies, comparing it to the supercar Ferrari F12 Berlinetta and the audiophile system B&W Nautilus. Indeed, this graphics card betrays the developer’s attention to every detail.
The face side of the PCB, which is 280 millimeters long, is covered by the cooler’s casing. Instead of trivial plastic or ordinary metal, the casing is made of aluminum coated with trivalent chromium and has a black magnesium-alloy insert around the fan and two polycarbonate windows through which you can see the GPU heatsinks. The whole thing looks very stylish. Indeed, the photo can’t convey the pleasure you feel while holding this card in your hands.
The reverse side of the PCB is exposed:
That’s okay because the exposed PCB is going to be cooled better than if it were completely sealed within a casing. There are highlighted letters “GEFORCE GTX” on one edge of the card. You’ll see it again in this review shortly.
There’s nothing on the other edge. We can note the lack of vent holes.
There is a vent grid in the card’s mounting bracket, though. You can also find three dual-link DVI-I connectors and one mini-DisplayPort here.
Thanks to this configuration of video interfaces, you can connect as many as four HD monitors to a single GeForce GTX 690.
There’s an opening in the opposite side of the casing through which you can see the fins of one of the GPU heatsinks.
Some of the hot air is exhausted through that opening into the system case.
The GeForce GTX 690 is equipped with one SLI connector, so you can build a multi-GPU configuration with two such cards and a total of four top-end GPUs!
At the back of the card we can see two 8-pin power connectors. The GeForce GTX 690 is specified to have a peak power draw of 300 watts and a typical power draw of 263 watts. A 650-watt or higher PSU is recommended for a computer with a GeForce GTX 690 inside.
The 10-layer PCB is populated by two GK104 Kepler chips, two 2GB sets of GDDR5 memory, a switch chip and power circuitry:
The power system contains 10 phases.
Both GK104 Kepler chips are revision A2.
If you’ve examined the specs, you must have noticed that the GeForce GTX 690 comes with full-featured Kepler GPUs that are not cut down in any way. The card has a total of 3072 unified shader processors, 256 texture-mapping units, and 64 raster operators. However, the base GPU clock rate is reduced in comparison with the GeForce GTX 680. It is 915 MHz in 3D applications or 91 MHz lower compared to the GTX 680. The boost clock rate is 1019 MHz, which is only 3.9% lower than the boost clock rate of the GeForce GTX 680. The GPU frequency is dropped to 324 MHz in 2D mode.
We couldn’t read the markings on the GDDR5 memory chips of our card but we suspect the GeForce GTX 690 to use the same memory as the GeForce GTX 680, i.e. FCBGA-packaged chips from Hynix Semiconductor labeled H5GQ2H24MFR R0C. There are 2 gigabytes of memory per each GPU. The card’s memory clock rate is 6008 MHz in 3D mode (the same as on the GTX 680) and 648 MHz in 2D mode.
GPU-Z 0.6.1, the latest version available as of our writing this, knows the GeForce GTX 690 well enough: