Before we proceed with temperature results, it is critical to note we have typically used FurMark in the past to test temperatures as a “worst case” scenario, but that causes significant throttling in many instances, and we no longer consider it a fair or accurate representation of usage or results, particularly with the new dynamic load adjustments on Kepler products.
Consequently, we switched awhile ago to using a “real world” scenario: we run Crysis 2 for 15 minutes and log the temperatures, reporting the highest heat output achieved. Crysis 2 loads up a card pretty heavily, but more importantly it’s also a very accurate representation of what a typical gamer might encounter in normal card usage as well. Quite simply, this approach avoids any claims of “unrealistic” or “biased” thermal testing; in our opinion, nothing is more realistic than actual gaming temperatures.
Lastly, note that we’re comparing cards with non-reference coolers below, which will make things even more challenging for the Nvidia reference card.
With that said, the results are below:
As we expected, the Gigabyte cooler is doing an excellent job at keeping the GTX 650 Ti very cool under load, peaking at only 53°C, a 9% improvement down to 5°C lower than the reference Nvidia design. The other positive to note here is the fan is noticeably quieter than the reference design as well. The fan here looks identical to those found on Gigabyte’s popular Windforce cooler, so it’s no surprise that this new cooler has such low noise levels as well.