Review: Silent power
KFA2 gained a lot of overclocking experience with the GTX 780 Hall of Fame (HOF), which we had a chance to check out a few months ago. The original GTX 780 HOF was based on the GK110 (GK110-300-A1) and KFA2 applied the lessons learned back than to its latest Hall of Fame card, the GTX 780 Ti HOF which is based on GK110 (GK110-425-B1).
We previewed the new card last week and we were impressed in more than one level. The design is quite nice and since it has a massive cooler it is quiet. The GTX 780 Ti HOF comes with a substantial factory overclock. KFA2 bumped up the base clock from 875MHz to 1020MHz, while the Boost clock stands at 1085MHz, up from 928MHz on Nvidia's reference design.
When we first tested the reference GTX 780 Ti we quickly realized that even its stock cooler offers exceptional performance. The stock cooler can deal with reference clocks or even a slight overclock, but the KFA2 card comes with a high factory overclock and it needs something a bit better. Note that the reference cooler can also handle a 145MHz overclock, but you have to trade silence for performance to get there. With a good custom cooler you don't - and KFA2 has a very promising custom cooler.
Some gamers like overdesigned gear, including coolers and shrouds. The HOF does not disappoint in the aesthetic department. The cooler looks mean, almost steampunkish and the white PCB is a nice touch, too. We're not at a beauty pageant though - all we want to know is how it handles the 145MHz factory overclock.
The next picture shows the KFA2 GTX 780 Ti which is just loosely based on the reference design.
Before we proceed to the actual results let’s refresh our memory. Nvidia launched the GK110 in May as the Geforce GTX 780 that came with 2304 CUDA cores, 192 texture mapping units (TMUs), 48 render output units (ROPs) and 3GB of GDDR5 memory running on a 384-bit memory bus.
Specification wise the GTX 780 ended up with 50% more CUDA cores than its predecessor, the GTX 680, and this among some other things gives the GTX 780 card a significant performance boost, although the GK110 is based on the Kepler architecture, also used in the GTX 680. The GK104 chip used in the GTX 680 has 1536 CUDA cores. The memory bandwidth was increased, too, thanks to the wider 384-bit bus. The GTX 780 has a 384-bit bus while the GTX 680 is limited to a 256-bit interface, but the memory speed on both cards is 6008MHz.The frame buffer was increased from 2048MB to 3072MB of GDDR5 memory. Compared to GTX Titan ($999), the GTX 780 debuted at a more attractive price of $649.
Nvidia’s response to AMD’s Hawaii was to slash the GTX 780 price by $150 and launch the more powerful Geforce GTX 780 Ti, which also rendered the GTX Titan obsolete.
The GTX 780 Ti has 25 percent more cores than the original GTX 780. The suggested retail price for the Geforce GTX 780 Ti was set at $699. The GK110-425-B1 that powers the GTX 780 Ti has 5 graphics processing clusters, 15 streaming multiprocessors, 2880 CUDA cores, 240 TMUs, 48 ROP units, and 3GB of GDDR5 memory running on a 384 memory bus. GTX 780 Ti card comes with faster memory 7000MHz effective, compared to the 6008MHz on the GTX 780.
Although the KFA2 GTX 780 Ti Hall Of Fame comes with a hefty factory overclock which raises the GPU clock by 145MHz, the memory is still running on reference clocks, i.e. 7000MHz (effective GDDR5). This is quite common on overclocked Nvidia cards today.