Not wanting to repeat what I’ve already said from the review of the GTX 780, the GTX 770 uses the exact same cooler ID as Titan, the only visual difference being the GTX 770 branding along the end of the cooler. In all other areas the cooler is identical to those seen on Titan and the 780.
Like the 780, the 770 has the updated fan controller which reduces fan speed fluctuations, in turn giving a more consistent acoustic output from the cooler.
Delving down into the core of the card, like the GTX 680, the 770 uses the GK 104 Kepler core, however as I’ve mentioned in the introduction, the GTX 680 cannot be flashed or modified in any way shape or form to turn it into a GTX 770. This is because the GK104 core we see here is a modified revision with more CUDA cores, texture units and the power system on the board is also different.
The memory on the GTX 770 is one of the biggest specifications that NVIDIA have to shout about, with a new series of ICs, the GTX 770 now comes as standard with a 7Gbps – 256-bit memory interface, a whole 1Gbps more than the 680. The 770 will also come in 2Gb and 4Gb versions and on the back of the PCB we can see where the additional GDDR5 ICs would be mounted.
Over the GTX 780, the GTX 770 has a slightly lower TDP of 230w, meaning that we find an 8 + 6 pin power connector layout on the end of the board.
SLI is a crucial highlight when it comes to boosting the performance of a single card, with up to 70% scaling to be seen within most gaming applications when set-up in a two -way configuration. For those wanting more, the GTX 770 will also support three and four way configurations.
The display outputs are also another specification that remains unchanged, with single and dual-link DVI, HDMI and DisplayPort 1.2.
The GK104 based card comes with a core clock speed of 1046MHz, boosting to 1085MHz through Turbo Boost 2.0 technology, a memory speed of 1753MHz (7Gbps effective), 1536 CUDA cores and a transistor count of 3.5b within the 28nm package.