For the 3DMark (2013) testing, we put the cards through two tests: Fire Strike and Fire Strike Extreme. Both consists of extreme levels of tessellation and volumetric illumination, as well as complex smoke simulation using compute shaders and dynamic particle illumination; while the latter will push the graphical processing capabilities of the cards further with more tessellation, more particle effects and more taxing DirectCompute calculations.
The higher overclocked Gigabyte card was the obvious winner and it outperformed the slower ASUS card by a thin margin of about 4% for the synthetic benchmark of 3DMark (2013). On average, they outperformed the reference GTX 780 Ti card by about 8%. Against the GTX Titan, the custom cards increased their lead to about13%. In comparison with their worthy AMD competitor, in the form of the Radeon R9 290X card, the custom GTX 780 Ti cards were in the lead by about 9%.
As a gauge of their in-game performance, we put the cards through the paces of Crysis 3. The game runs on the CryEngine 3, with extreme amounts of tessellation, per-pixel per-object motion blur, Bokeh Depth of Field, displacement mapping on small terrain, particle and volumetric lighting and fog shadows, dynamic cloth and vegetation, caustics and diffuse shadows. The average frame rates generated during a fixed, rendered cutscene, is recorded to gauge each card's performance.
For both cards, their leading margins seen in 3DMark (2013) were generally widened across the board in Crysis 3. Pitted against each other, the Gigabyte card was about 2- to 4% better than its ASUS counterpart. Against the rest of the older NVIDIA cards, their margins were widened slightly, from about 9% against the reference GTX 780 Ti, to roughly 19% against the Titan. Against the GTX 780 card, the custom 780 Ti pair were about 24% better. The AMD Radeon R9 290X was no match as well as the custom NVIDIA pair outperformed it by 32% on average.
For our overclocking exercise, although we managed to achieve higher operating frequencies for the Gigabyte card, it was the overclocked ASUS card that pulled ahead marginally. To reiterate, the new GTX 780 Ti cards support both GPU Boost 2.0 and the power balancing feature that is currently exclusive to the GTX 780 Ti. Therefore, we believe as a result, the overclocked GTX 780 Ti cards achieve a somewhat similar performance levels as both these technologies attempt to maximize each card's performance level. As seen from the graphs below, all three GeForce GTX 780 Ti cards (including the reference card) managed very similar overclocking performance.
Having said that, for those interested, we calculated the performance in percentage gain from overclocking for each card just for kicks:- the reference GTX 780 Ti card achieved the highest gain of about 20.6% in comparison to its normal operating state, ASUS GTX 780 Ti card managed a 13.6% increment, and the Gigabyte card was third with its 8.3% gain. Obviously, the figures differ only because each card has a different base level of performance, but it also tells you if it's worthwhile overclocking it. In this case, if you have a reference clocked GeForce GTX 780 Ti, it makes a lot of sense to push it further. On the other hand, the need to overclock has diminishing returns for the speedier custom cooled cards as they top out at about the same performance levels as a reference-designed card.
By default, all the new GeForce GTX 700 series cards have a default temperature threshold of 80 degrees Celsius to maximize upon. This is because GPU Boost 2.0 factors temperature to boost the card's operating clocks if there's available thermal, power budget to consider (you can refer to our previous write-up on how it works). Having said that, it seems that the coolers on both the ASUS and Gigabyte cards are so efficient that even with the increased operating clock speeds, the operating temperatures of these cards were only hovering about 60 degrees Celsius - at load levels!
Between the ASUS and Gigabyte cards, the latter has a much more efficient cooling system as it has higher operating clock speeds to tame, but it registered temperatures no higher than the ASUS card.
The power consumption figures of both cards are rather interesting because it can either be said that Gigabyte is very power efficient or ASUS is drawing more power than expected.
Both cards turned in impressive performances and it was the Gigabyte card take took the lead on most occasions due to its higher overclocked graphics core. Further to that, the Gigabyte GTX 780 Ti Windforce 3X OC card had a lower power draw than the ASUS card and a more efficient cooler. As such, it had a very compelling package.
ASUS' offering is still good, but it just wasn't better than the Gigabyte counterpart. There's certainly nothing wrong with the ASUS card and we would wholeheartedly recommend it if you prefer a heavy-duty look to your graphics card. About the only aspect we found puzzling was its much higher power draw and we can't help but wonder if it had anything to do with the implementation of a pair of 8-pin power connectors. Perhaps ASUS thought more power draw would be advantageous for better overclocking, in light of the new power balancing technology of the GTX 780 Ti. In reality as we found out, it had almost no impact for the actual overclocked performance. If you have plans to disregard its official cooler and get down to modding your own high performance water cooling block, perhaps the extra power might come in handy to try and push the card closer to its limits, but that's a lot of effort for something that already performs rather well out of the box.
Overall, it is evident that the Gigabyte GeForce GTX 780 Ti Windforce 3x OC card is the better candidate despite the strong showing from the ASUS card. The ASUS GeForce GTX 780 Ti DirectCU II OC 3GB GDDR 5has a suggested retail price of S$1,119, but we've found its retail price to be lower at S$1,089. Gigabyte's card was also found to be similarly priced at about S$1,069. Clearly, the Gigabyte card manages to edge out its ASUS counterpart with slightly better value. However at these four-figure price points, the differences are negligible and you would be better served to consider which card's build better fits the design elements of your rig.