Cooling, Noise, and Power
I think the most important tests are the tests that let us look at what sets one brand apart from another and from card to card from the same company with the same GPU. For the most part, all of the other tests are predetermined by the GPU used, the clock speed they set it too, and with some cards the memory capacity. Of course, thermal throttling can play a role too. So checking out power usage, cooling, and noise is important to help you pick which model of card you want once you decide which GPU you want. It is especially important for small form factor builds where all three become even bigger issues.
My first set of tests on the RTX 2070 XC Gaming were to look at its power usage. To do this I put the test bench under a load and document the peak wattage pulled from the entire system with our Kill-A-Watt. So keep in mind these numbers include the CPU, SSD, and cooling and are not just the GPU. But when compared together still give a good idea of the power usage you can expect. Anyhow, I tested with two different loads. First I used 3DMark Fire Strike in its combined benchmark that loads the CPU and GPU. This is close to what you can expect with regular in game usage (adding RTX or DLSS usage could increase this). The EVGA card came in 4 watts less than our Founders Edition card and more importantly 29 watts less than the overclocked GTX 1080 that was somewhat close in performance. The new RX590 is right up with the 2070 as well when they aren’t in the same performance class. My second test was using ADIA64’s GPU stress test, putting only load on the GPU but a more synthetic and demanding load on the GPU than most games. Here the EVGA XC pulled 263 watts and the Founders Edition pulled 270. That RX590 was even higher here and I was surprised at how close the slower clocked Gigabyte RTX 2070 was to the XC.
Next, I wanted to check out noise levels of the two EVGA logo covered fans. My second graph shows the RPM that the fans run when turned up to 100% fan speed. I keep that for reference. Then the noise graph shows decibel readings from 18 inches away at a 45-degree angle from the fan side on our open-air test bench. You can expect lower numbers in a case, but this allows me to get accurate numbers when comparing card to card. I test at 100% and 50% fan speeds for consistency. What I found was the RTX 2070 XC was right in the middle of the pack. Quieter than the new dual fan Founders Edition cards but a little louder than some of the single fan cards. This was also about where the XC fell in the fan speed charts as well, so nothing really out of the ordinary there.
My last round of testing was taking a look at cooling performance. To me, this is the most important of all of the card specific tests. Not only does better cooling allow for more overclocking or better clock speeds, but it also means the fans may run at a lower RPM on average keeping even a loud card quiet. I ran this test twice as well, both times using AIDA64’s GPU stress test to load up the GPU. I ran the test for a half hour or longer if the temps still hadn’t leveled off (they normally are). Then I tested with the stock fan profile and then again with the fans turned all the way up. This way we can see the expected performance and then the potential cooling performance. The delta between the two can mean more overclocking room if you are willing to put up with the fans cranked up. This is where the EVGA RTX 2070 XC performed the best. It was the second coolest card in our stock fan settings test. That Asus RTX 2080 was still out ahead with its huge cooler. Then at 100% fan speed, it dropped down to 47 degrees from the 61 before. The Gigabyte RTX 2070 with a lower clock speed and three fans did perform better in the second test, as did the water-cooled Vega 64.
While I was waiting on the temperature testing I did also get a few thermal images of the EVGA RTX 2070 XC. It gives us a good look at how the card blows its heated air out the top and bottom of the card, especially with some of that heat trapped under the card. Like with any other GPU like this I wouldn’t recommend mounting the card over an M.2 drive. The openings in the backplate seems to be doing a better job of letting that heat out on the back of the card. You can see each opening is warmer than the fully covered areas.