EVGA has a strong foothold in the custom PC market with products ranging from hardware to accessories. They are a well-known, respected, and trusted manufacturer that any enthusiast should be familiar with. To that end, EVGA has been kind enough to allow us to test their top RTX 2060 offering, the XC Ultra Gaming.
The EVGA RTX 2060 XC Ultra uses a TU106 GPU that’s cut back by 16.67% (the RTX 2070 has the full TU106 GPU). The GPU has a base clock of 1365 MHz and a rated boost clock of 1830 MHz. As for memory, there’s 6 GB of GDDR6 running on a 192-bit bus that’s clocked at 1750 MHz. The rated TDP is pretty tame at 160 W which means rather small PSU can power a system with the EVGA RTX 2060 XC Ultra.
If you’re in need of a features refresh for Turing, head on over to our RTX 2080 and RTX 2080 Ti review where Joe covered that material in detail.
|EVGA RTX 2060 XC Ultra Specifications|
|GPU Base Clock||1365 MHz|
|GPU Boost Clock||1830 MHz (1950 MHz actual)|
|Frame Buffer||6 GB GDDR6|
|Memory Clock (Data Rate)||1750 MHz (14 Gbps)|
|Memory Bandwidth||336 GB/s|
|Render Output Units (ROPs)||48|
|Texture Mapping Units (TMUs)||120|
|Texture Fillrate||219.6 GT/s|
|Pixel Fillrate||87.8 GP/s|
|GigaRays /sec||5 GR/s|
Here’s a GPU-Z screenshot confirming specifications…
A key feature of EVGA’s RTX 2060 lineup is the ability to choose between a longer, dual fan, 2 slot card or a shorter, single fan, 2.75 slot card. Having a 2.75 slot card is nice for cooling, but I was a little disappointed that it wouldn’t be a good fit for SFF ITX builds since most of those cases would have a max of 2 slots available. Anyways, the card we have today, the XC Ultra, is one of the longer 2 slot cards.
The new hydrodynamic bearing fans are supposed to reduce noise by 15-19% when compared to the more traditional sleeve bearing fans.
EVGA has also redesigned their signature Precision X1 software to make overclocking, fan profile customization, monitoring, etc. as easy and intuitive as possible.
Packaging & Accessories
The front of the box shows a picture of the RTX 2060 XC Ultra with the model name to the left of the image. EVGA has their branding in the upper left corner. The bottom third has NVIDIA’s GeForce RTX branding along with the 2060 model at the bottom edge. The back of the box lists a few features in six languages. The bottom third of the back mentions EVGA’s Precision X1 and NVIDIA’s GeForce Experience software.
Inside the box, the RTX 2060 XC Ultra is wrapped and padded in two layers: one layer of typical shipping bubble wrap and one layer of black foam. The packaging definitely isn’t as nice as NVIDIA’s, but it gets the job done just fine. As far as accessories go, they’re basically non-existent with only an Installation Guide pamphlet included with the video card.
EVGA GeForce RTX 2060 XC Ultra
The EVGA’s RTX 2060 XC Ultra has an all black, dual fan shroud covering a large heatsink. EVGA’s branding is scattered around the shroud between the fans, on the edges of the shroud, and even on the fans themselves. The style is what I would describe as aggressive; consisting of hard lines and sharp angles as opposed to smooth curves.
Looking at the back of the card we can see that the PCB doesn’t span the full length, and there is a vented backplate attached to the end of the heatsink where there’s no PCB.
Here are a couple of isometric shots to further showcase the design and style of the card. On the top of the card, EVGA has their branding and GPU model across the length which could be seen through a windowed side panel of the case.
A Closer Look
The video output consists of DVI-D, HDMI, and two DisplayPorts which should satisfy the vast majority of users. However, compared to NVIDIA’s Founders Edition and some other Turing-based GPUs, there is a missing USB Type-C output which would be used for virtual reality purposes. A single 8-pin PCI-E power connector supplies the RTX 2060 XC Ultra with all the juice it needs. The power connector is located in the usual orientation on the end of the PCB, which isn’t the same as the end of the whole card since the PCB is short.
After removing the heatsink from the EVGA RTX 2060 XC Ultra, there seems to be gratuitous amounts of thermal paste applied to the GPU. EVGA made sure to get full TIM coverage across the TU106, so much so that it spills over the edges of the GPU.
Since the GPU’s VRM is one of the hottest areas of the card, its thermal pad spans the heatsink’s three heatpipes and contacts them directly.
The heatsink’s fans use a single plug that connects to the end of the PCB. We can also see in this photo the partial backplate covering the end of the heatsink that extends past the PCB. Looking closely at the fans themselves we can see the fan blades were manufactured with EVGA’s signature “E” logo covering the blades. I’m unsure whether this will help or hinder the fans’ performance in any noticeable way.
The thin aluminum plate covering most of the PCB acts as a heat spreader for the GDDR6 memory while using thermal pads to make contact.
After cleaning up the thermal paste, we can see the Turing TU106-200A GPU. The full TU106 would be the TU106-400 which is used in the RTX 2070, so the “200” means this is the trimmed chip. The “A” means the the GPU is binned to allow factory overclocking.
The VRAM is comprised of six 1 GB Micron D9WCW chips rated at 14 Gbps and clocked at 1750 MHz.
The overall power delivery is a 6+2 setup with the GPU using six phases and the VRAM using two phases. Power is supplied to the card via the PCIe slot (75 W) and a single 8-pin PCIe power connector (150 W) for a total of up to 225 W.
Here’s a picture of the RTX 2060 XC Ultra installed on my test bench
Test System & Methodology
Our test system is based on the latest mainstream Intel platform, Z370, and uses the i7-8700K 6c/12t CPU. The CPU is overclocked to 4.7 GHz on all cores/threads with cache set to 4.3 GHz. The clock speeds used provide a good base to minimize any limitations the CPU may have on our titles, particularly when using the lower resolutions, and should be attainable with a good air cooler or better. DRAM is in a 2 x 8 GB configuration at 3200 MHz with CL15-15-15-35-2T timings which is a middle of the road option that balances performance and cost.
|CPU||Intel i7 8700K @ 4.7 GHz, Cache @ 4.3 GHz|
|CPU Cooler||EVGA CLC 240|
|Motherboard||ASUS ROG Strix Z370-F Gaming|
|RAM||2 x 8 GB G.Skill TridentZ DDR4-3200 15-15-15-35 (1.35 V)|
|GPU||EVGA GeForce RTX 2060 XC Ultra (417.71 Drivers)|
|Storage||256 GB Transcend MTS400|
256 GB OCZ Vertex 4
|PSU||Seasonic X-750 Gold|
|Fluke 52 II Dual Input Thermometer|
|Tenma Sound Level Meter|
We have made some significant changes since the last update, adding a few new titles and dropping some of the older games. More details can be found in the GPU Testing Procedures article which has been updated with our latest benchmarks.
Here’s a brief rundown…
- GPU performance testing is done according to our video card testing procedures.
- Peak system power consumption was measured using a Kill-A-Watt meter during both Shadow of the Tomb Raider and F1 2018 (5 laps).
- For temperature testing, the GPU is stressed with Shadow of the Tomb Raider and F1 2018 (5 laps) at each 10% fan speed increment from 20-100% and peak GPU temperatures are logged with GPU-Z.
- For cooler performance testing, the GPU was stressed with ~20 minutes of looping 3DMark Time Spy for each 10% increment in fan speed between 20-100%. Peak GPU temperatures are logged with GPU-Z.
- For noise testing, all external sources of noise are removed or turned off if possible (TV, ceiling fans, A/C, etc.). Then, sound level is measured at 10 cm (~4 in) from the intake side of the card and fan speed is manually varied using 10% increments from 20-100%.
- All testing was performed on an open bench before disassembling the graphics card.
For synthetic tests we stick with Underwriters Laboratories (UL) 3DMark, specifically the Fire Strike Extreme and Time Spy tests. Fire Strike Extreme runs at the common 1920×1080 (FHD) resolution while Time Spy runs at 2560×1440 (QHD). EVGA’s RTX 2060 XC Ultra performs above NVIDIA’s RTX 2060 by around 4-5% in both tests due to its higher GPU boost clock. It also beats out the Vega 64 in the higher resolution Time Spy benchmark.
These benchmarks are always done in 1920×1080 (FHD) since it’s easily the most common resolution. According to the January 2019 Steam Hardware Survey, ~90% of primary resolutions are 1920×1080 or below. However, testing at 2560×1440 (QHD) is done as well to stress the GPUs and help showcase their potential.
The RTX 2060 XC Ultra performs very similar to NVIDIA’s RTX 2060 and AMD’s RX Vega 64 in F1 2018. EVGA’s RTX 2060 comes out ahead in World of Tanks by a noticeable amount over the Vega 64 and slightly above NVIDIA’s RTX 2060.
There’s a little different story for the Vega 64 in Far Cry 5 and The Division. It comes out ahead of the RTX 2060s in both titles, while EVGA’s RTX 2060 XC Ultra again tops NVIDIA’s by a small margin.
I know it’s starting to sound like a broken record, but the EVGA RTX 2060 XC Ultra outperforms NVIDIA’s yet again in both Final Fantasy XV and Shadow of the Tomb Raider. Interestingly the Vega 64 was rather horrible at the Final Fantasy XV benchmark, definitely performing below what would be expected.
In Ashes of the Singularity, the RTX 2060s and the Vega 64 are on par with each other with only a FPS or two separating them.
When turning up the resolution, we can see the EVGA RTX 2060 XC Ultra fall below the magic 60 FPS mark in Ashes of the Singularity and just miss the mark in Final Fantasy XV by 0.1 FPS. It falls down to 62 FPS in Shadow of the Tomb Raider at this resolution and stays well above the 60 FPS mark in the other titles. EVGA’s RTX 2060 XC Ultra stays slightly above NVIDIA’s RTX 2060 across the board and trades blows with AMD’s RX Vega 64 as well.
Cooling & Noise
The first round of temperature testing is just keeping the fan profile at its default settings and running through a couple of benchmarks. The EVGA RTX 2060 XC Ultra was the coolest card of the bunch and didn’t break 70 °C in either Shadow of the Tomb Raider or F1 2018.
Next up I wanted to test out the performance of EVGA’s cooler across all fan speeds to see if there was an obvious optimal fan speed to balance temperatures and speed. Here, I used looping 3DMark Time Spy for ~20 minutes at each fan speed increment to make sure the GPU leveled out in max temps before recording. After plotting the data, it looks like there is a pronounced inflection (at around 50% fan speed) where a 30% increase (from 20% to 50%) gave us 23 °C cooler temps, but going further from 50% to 100% only gives an additional 10 °C. So, keeping the fan around 50% will result in the best temps per fan speed. But…what about noise?
For noise or sound pressure level (SPL), I had to measure close to the intake (10 cm) to get an accurate reading within my meter’s range. Then, the other distances were estimated using the following:
L2 = L1 – 20 * log10(r2/r1)
- L1 = Sound level at the reference distance
- L2 = Sound level at the desired distance
- r1 = Reference distance
- r2 = Desired distance
One note I’d like to make is that typical linear percentages cannot be used when comparing dBA measurements because dBA uses a logarithmic scale. So, a “rule-of-thumb” is every 1 dBA difference is roughly equal to a 10% difference in perceived loudness.
On to the results, 20-30% fan speed was basically the same as the ambient sound level which can be seen from the relatively horizontal plots between 20% and 30% fan speed. Looking at the actual recorded data (blue line), once past the 30% mark, the dBA reading rose quickly at 40% fan speed going from 21.5 dBA to 39 dBA. After 40% fan speed, the dBA reading rose pretty steadily and almost linearly by around +10 dBA every +20% fan speed.
Now, most people won’t be sitting 4″ or even 12″ away from their PC, so let’s look at the 24″ to 60″ plots for a more real-world estimate. Remember in the previous section that around 50% fan speed seemed to net the best temps for the speed. If we look at around 50% here, we see that those plots sit between 20-30 dBA and that even moving up to 60% fan speed doesn’t cause much of a jump in noise. So, having the fan speed max out at somewhere around 50% will provide the best balance in temps and noise/speed. Although, there’s plenty of temperature headroom to lower the fan speed if desired.
The EVGA RTX 2060 XC Ultra had the lowest idle power consumption among the cards tested, but reached at or above RTX 2070 power consumption when loaded with Shadow of the Tomb Raider and F1 2018. This isn’t much of a surprise since the XC Ultra is a manufacturer overclocked model.
Here are a couple of shots of EVGA’s Precision X1 software showcasing the sections I used most. GPU clock, memory clock, voltage, power target, temperature target, and fan speed can be adjusted on the main home page of the software. When cycling through sections in the bottom half of the window using the arrows, you’ll come to a section for scanning and testing overclocks. Of course, there are other sections like fan profile adjustment, overall profile saving, LED manipulation, hardware monitoring, and on-screen display setup.
First off I tested the stock settings to see what actual clocks the GPU reached during load. The advertised boost clock on the RTX 2060 XC Ultra is 1830 MHz, but the GPU actually boosts to 1950 MHz. I noticed the memory didn’t run at the rated 1750 MHz, even though GPU-Z showed that as the default memory clock. When loaded at stock settings, the memory stayed a little lower at 1700 MHz instead.
Overclocking was done by first using EVGA’s Precision X1 OC Scanner to auto overclock for us which resulted in +92 on the GPU. From there I started with +100 on the GPU to tested and went up by +10 increments after every passed test, then tried dropping down by -5 after a failed test, driver error, or whatever type of system crash. I ended up with +130 on the GPU which resulted in a 2085 MHz GPU clock that even spiked to 2115 MHz briefly (see GPU-Z screenshot).
For the memory, we used increments starting at +1000 (+250 MHz) and failing, then backing down to +800 to pass. Then, I made my way up to and settling at +900 (+225 MHz) which gave me a 1925 MHz clock that spiked to 1975 MHz (see GPU-Z screenshot).
The overclocked Time Spy run at 8596 is a ~7% gain in performance and only 5.3% lower than the stock NVIDIA RTX 2070 run. Of course, the RTX 2070 can be overclocked as well to widen the gap between the two. The GPU temperatures only got to 68 °C and fan speed never exceed 49% (using default fan profile).
At stock, the performance of the EVGA RTX 2060 XC Ultra landed where we expected, somewhere between the standard NVIDIA RTX 2060 and the RTX 2070. When overclocked, the RTX 2060 XC Ultra gained ~7% and came much closer to the stock RTX 2070 results. The large heatsink and dual hydrodynamic bearing fans kept the RTX 2060 XC Ultra nice and cool throughout testing never hitting 70 °C even while overclocked. Fan noise wasn’t noticeable until hitting around 45%+ on the fan speed even when using my open test bench.
EVGA’s Precision X1 made fan speed adjustment, overclocking, and testing a breeze. There are many features outside of the ones used for this review that really makes for an all-inclusive software package.
EVGA’s RTX 2060 XC Ultra comes in at $379.99 (NewEgg, EVGA), which is $30 above NVIDIA’s $349.99 MSRP. In my opinion, the markup is worth it for cooling alone. The RTX 2060, in general, is priced too high for me to recommend for a true mid-range build, but the performance you get for the money from the RTX 2060 is hard to beat.
Overall, the EVGA RTX 2060 XC Ultra performs well at stock and overclocked and has a great cooling solution for both performance and low noise. It’s definitely a card I’d recommend for those looking to get the most out of their money.
– Matt Green (MattNo5ss)