NVIDIA has undoubtedly ruled the computing graphics market. The last two years have been a perfect storm as PC gaming has enjoyed a huge resurgence, and people who might have otherwise overlooked the segment found a purpose – thanks in no small part to the cryptocurrency boom. The 10 Series graphic cards from 2016 is their best-selling hardware lineup to date, but the world – and eager fans – seem divided on where GPU innovation is headed next.
For this, NVIDIA has unleashed two new contenders for your precious dollars: the GeForce RTX 2080 Ti and GeForce 2080 RTX. This review focuses on the latter, which isn’t just more affordable but a decent choice if you’re just looking for an upgrade to an existing system. If you’re just after a more pristine gaming experience than either card will make your current rig sing, but those on the fence about which card to pick up should know what set the two apart.
First off, the GeForce RTX 2080 is the standard flagship GPU featuring the company’s latest Turing architecture. It may surprise some, but I have a more receptive opinion about this card versus its pricier option. As a matter of fact, I probably spent a little more time with this card than the latter just to see what’s exactly lost by saving a few hundred bucks (quick answer: not much). However, lingering criticisms such as missing technological advancements in ray-tracing, DLSS, and efficiency are still valid; just some FYI.
Tentative Technology
Ray-tracing is the biggest capability NVIDIA wants buyers to concentrate on, with claims that this technology is supposed to be the ‘holy grail’ in graphical prowess. This means realistically simulating the lighting of a scene and its objects by way of hybrid implementation — with accompanying rasterization techniques for rendering physically correct reflections, refractions, shadows, and indirect lighting in real-time. This was something deemed impossible for modern graphics cards to achieve – until now.
Deep Learning Super-Sampling (DLSS) is debatably the more attractive feature. It’s a new form of antialiasing (AA) that — in practice — utilizes AI-learning and auxiliary RTX tensor cores. This technique that produces a smoother and incredibly detailed imaging effect if an algorithm is co-developed for specific titles.
In order words: it’s supposed to exceed current antialiasing methods (TAA, FXAA, SMAA, etc.) by scanning PC games allowed by the developer, and calculations being remotely distributed by a supercomputer back to the end-user. A performance boost of 30%-45% is theoretically possible, reducing the needed amount of rendering power and physical resources involved.
Sleeker and Cooler
Like the RTX 2080 Ti, NVIDIA addresses numerous criticisms with older Founders Editions. The OEM versions are meant to be centerpieces of how their own cards would perfectly appear, with an understated design that fits within any serious machine. However, the exterior functionality of previous cards weren’t great when it came to cooling, due to the reliance of a single (HSF) blower fan.
Now, that old layout is ditched and finally adopts a dual blower setup that breathes easier and runs a lot more quietly. The external appearance also scales back the flashiness for subdued practicality, with an aluminum backplate that’s less flamboyant and functional to dispel heat. One trait that is faithfully retained is the illuminated outward-top side that proudly displays the name, although now it says “GeForce RTX”.
All-Connectivity
Connectivity incorporates one HDMI (2.0b) and three DisplayPort (1.4a) ports with HDCP 2.2 DRM encryption, the latter outputs theoretically able to handle up to 8K (7680×4320) of pixel resolution through a single DSC 1.2 cable. VR-ready users are also accounted for with an additional VirtualLink USB-C connector meant for next-generation virtual reality headsets. Legacy users aren’t completely left out as a DP-to-DVI adapter is included in the box.
Specs and Power
Compared to Pascal-based GeForce/TITAN X models, the RTX 2080 is a mild step-up when it comes to technical figures. The reference specs for RTX 2080 Ti are plentiful with 2944 CUDA cores, 1515MHz base/1710MHz (1800OC) boost clocks, Micron 8GB GDDR6/14Gbps, and memory bandwidth of 448GB/s with 256-bit interface. What is unique is the addition of 60 trillion (FE) or 57 trillion (reference) with RTX-OPS at 8 Giga Rays/s, which is solely provisioned for ray-tracing performance.
Take note that overall figures appear to be higher than the 2080 Ti, possibly to balance imposed bandwidth and bit interface limitations.
Another thing to note is required power, which the RTX 2080 needs quite a bit to run properly. In fact, NVIDIA recommends you have a 650W minimum PSU because a 14 ATX pin (6 pin + 8 pin) connection must be connected in order to boot. For example: something like an EVGA SuperNOVA 1000 G+ or minimum 800W power supply is ideal. Standard wattage at initial load is 260 watts and idling draws a noticeable 45 watts of power consumption, that’s a lot of juice although there are plans on getting the voltage down later on.
Performance Testing
The RTX is being touted as a de-facto choice for both 4K gaming and HDR fidelity, whether you have a benchmark DIY machine or wanting a legitimate alternative to regular home entertainment arrangements. For this, I borrowed both a 27-inch ASUS ROG Swift PG27UQ monitor and Epson Pro Cinema LS10000 4Ke Laser Projector as these displays could properly accentuate the demands of current AAA titles. As you can imagine, this is expensive gear largely reserved for the PC/AV elites.
Test System
- AMD Ryzen 7 1800X Desktop Processor
- MSI X370 XPOWER GAMING TITANIUM AM4 Motherboard
- Crucial Ballistix Sport 32GB DDR4-2400 Dual-channel DIMM
- Toshiba RC100 240GB M.2 SSD
- Crucial MX500 1TB SSD
- EVGA SuperNOVA 1000 G+ Power Supply
- InWin Polaris RGB Dual Case Fans
The RTX 2080 is immensely suited for the standard 1440p/144Hz in today’s gaming circles, and I had no problem getting this card to run at maximum settings while keeping a consistent rateframe throughout testing. It matches and often exceeds the capability of former GTX 1080 Ti/TITAN Xp heavyweights. When compared on dollar/performance basis the RTX 2080 is an upgrade.
I tested a number of games using Shadow of the Tomb Raider, Call of Duty: Black Ops 4, Final Fantasy XV, DOOM and Far Cry 5. Technical demonstrations of Star Wars Reflections DLSS Demo, Final Fantasy XV Benchmark, and Infiltrator DLSS Demo from Epic Games’ were also provided to experience the RTX in action under controlled elements.
To answer the blazing question everyone is thinking, both RTX models can handle 4K PC gaming and trounces its predecessors by a decent margin…but for whatever reason still inadequate for graphically intensive titles recently released. In fact, only a few of the games were able to effectively hit and maintain 4K/60fps minimum at the default highest (non-customized) presets, let alone reach the coveted 144Hz value.
Only the RTX 2080 Ti is more appropriate for the task, as the regular RTX 2080 matches the GTX 1080 Ti and certainly won’t cut it without some compromise.
Of course, these deficiencies may revolve around the current lack of ray-tracing and DLSS implementation, which weren’t available at the time of this review. Incredibly, Tomb Raider is supposed to be a flagship game for benchmarking and pushes the 2080 more than expected at an average 34fps at 4K/60Hz. No matter what I tried, the numbers seldom reached 50fps at best, and even this was a rare occurrence and usually maxed out 41fps.
Final Fantasy XV is another title being pushed by Square Enix for RTX 4K DLSS, but only through a press benchmark demo. Regardless, the DLSS showcase topped at around 53fps while actual in-game figures maxed out at 36fps. The 2080 fared better when playing Far Cry 5 at an average of 59-64fps, 57-63fps in CoD:BO4, and 134fps in DOOM (Vulkan API).
The other RTX video demos such as Infiltrator, had the 2080 start at 78fps then gradually level to 63fps in 1440p, 4K results were slightly lower results at 69fps/54fps respectively. Meanwhile, the entertaining Star Wars Reflections real-time DLSS video was locked at 24fps but looked great whether it was viewed in 2560×1440 or 3840×2160.
The takeaway is that both consumer-grade cards appear to be comfortable processing monster workload combinations in soft shadows, reflective lighting, and hue intensity all at once. These are controlled examples, but considering the visual enhancements with no graphical penalty these cards show the room for improvement.
Conclusions
Similar to the RTX 2080 Ti reviewed earlier, the GeForce RTX 2080 has a lot going for it on paper. Since its global debut at Gamescom, there’s been concern and growing backlash that the RTX series is more proof-of-concept than a substantial upgrade from previous iterations. These assumptions aren’t entirely off base by the fact that people are eagerly picking the current GTX lineup and getting much of the performance they want right now.
I was comparatively critical towards the 2080 Ti because you really don’t get a whole lot more real-world performance before overclocking — that card in essence is a guilty pleasure for well over a thousand dollars. The mesmerizing inclusion of ray-tracing technology and DLSS is locked away, and the actual rollout won’t happen until sometime next month at the earliest. Judging from the demos on-hand though, there is much to be excited about — everything just seems to be in a strange place for Turing GPUs at the moment.
Still, the RTX 2080 is a beast (it handily outperforms the 1080 Ti/TITAN Xp) while still managing to be a practical solution to most of your future-proofed 4k gaming needs. NVIDIA just needs to get things fully off the ground and release the necessary software updates to really show off its capabilities, which I’m sure will be impressive.Truthfully, neither RTX card is what you’d call “affordable” by any definition, but if you’re salivating over the potential of graphical goodness, you really can’t get much better right now.