On Tuesday, Nvidia Super-fied its GeForce GTX 16-series graphics cards, with the mass-market upgrade following in the footsteps of its beefier RTX Super siblings—and running defense against AMD’s impending Radeon RX 5500 series. The GeForce GTX 1660 Super launches today (we’ve already reviewed it), bolstered by some killer new software features in GeForce Experience, and a GTX 1650 Super is waiting in the wings in November.
Let’s start with the new hardware before diving into the software improvements.
The $229 GeForce GTX 1660 Super largely mirrors the specifications of the spectacular original GTX 1660: clock speeds, CUDA counts, texture units, die size—the underlying graphics processor remains identical in the Super.
The key difference? Memory. Nvidia swapped out the slower GDDR5 memory of the original GTX 1660 for cutting-edge GDDR6 VRAM in the new GTX 1660 Super, and oh, what a difference it makes. In our testing across a variety of games, the upgraded memory pushes the new card well beyond the original GTX 1660’s performance and closer to the frame rates pumped out by the $280 GTX 1660 Ti.
Again, check out our review for more details, but the Super version is well worth the extra $10 over the original GTX 1660’s price.
Make that the original GTX 1660’s original price, rather. While the vanilla RTX 2070 and 2080 disappeared when their Super-fied variants arrived, the original GTX 1660 is sticking around “at a reduced price point.” Nvidia wouldn’t provide any official pricing information, but if you poke around at online retailers, you should be able to find plenty of ostensibly $220+ GTX 1660 boards selling for around $200 to $210, sometimes after rebate. Don’t be surprised to see the original model’s new pricing settle in around there.
And yes, Nvidia’s reshuffling means that it now sells a GeForce GTX 1660, a GTX 1660 Super, and a GTX 1660 Ti. That’s not confusing at all, eh?
Slotting in below that will be the new GeForce GTX 1650 Super, which will receive a more substantial overhaul when it launches November 22. The original $150 GTX 1650 (which is also sticking around) offers minimal appeal unless you’re using it to upgrade a prebuilt non-gaming PC. It runs slower than the rival Radeon RX 570, and it costs more money—but unlike AMD’s GPU, many models of the original GTX 1650 don’t need extra power cabling (they draw all their power from their motherboard slot).
The GTX 1650 Super won’t be able to pull off the same trick. In addition to a GDDR6 memory upgrade, Nvidia’s overhauling the core GPU at the heart of its entry-level GPU. While the original GTX 1650 uses Nvidia’s smallest TU117 GPU, the new GTX 1650 Super uses a cut-down version of the larger TU116 GPU found in the GTX 1660 series. It should be significantly more powerful, with 1,280 CUDA cores (compared to the original’s 896), higher clock speeds, the aforementioned 4GB of upgraded GDDR6 memory over a 192-bit connection, and well, just more everything.
It also includes Nvidia’s newer and much more efficient “Turing” Encoder, after the company received criticism over saddling the original GTX 1650 with its last-gen encoder. “Turing NVENC is up to 15 percent more efficient—requiring 15 percent less bitrate at the same quality level—than previous-generation Pascal NVENC when encoding with H.264, and 25 percent for HEVC,” Nvidia’s reviewer's guide says. “In other words, you get an instant upgrade in image quality without having to bump up your streaming rate.”
EposVox, an excellent source for streaming reviews and information, tested Turing NVENC in RTX GPUs and called it “beyond impressive.” The enhancements boost the GTX 1650 Super’s total power up to at least 100W, however, so it can’t be powered solely by the motherboard.
The one question left? Pricing. Nvidia says it’ll reveal that information closer to the card’s November 22 launch. It’s probably playing coy to give the company room to counter-punch potential AMD pricing shenanigans around the Radeon RX 5500 series, after the “jebaited” controversy around the Radeon RX 5700 series release.
Nvidia won’t release its own Founders Edition graphics cards for the new GTX 16-series Super GPUs. Instead, you’ll find custom models by board partners like Asus, EVGA, and Gigabyte on store shelves. As a reminder, the GTX 16-series GPUs do not include dedicated real-time ray tracing hardware like their Turing GPU-based siblings, the pricier GeForce RTX 20-series.
GeForce Experience gets new features
You may have heard of ReShade. The enthusiast software lets you apply post-processing shaders to games, giving them a new look and feel, sometimes with jaw-dropping results. Now, Nvidia’s baking it into GFE, letting you use hundreds of filters in over 650 games.
There are limitations: Only select “official” filters work in competitive games to prevent you from being banned, and games that haven’t integrated Ansel at a developer level have their support capped to official filters as well. The 80-plus games that bake in the Ansel SDK can use any ReShade filter when the game is paused. More than 160 non-competitive games can apply any ReShade filter using Nvidia’s real-time Freestyle post-processing.
Get your screenshotting fingers ready, Dead End Thrills wannabes. This is huge for “games as art” aficionados.
To use ReShade filters, you’ll need to drop them into the folder at Program files > Nvidia Corporation > Ansel. (If you don’t see the folder there, you’ll need to create it.) Once that’s done, you’ll simply need to press either Alt + F2 to activate Ansel or Alt + F3 to activate Freestyle, then select your chosen ReShade option in the drop-down Filters menu. Easy-peasy.
Speaking of, the revamped image sharpening feature Nvidia introduced to combat AMD’s Radeon Image Sharpening once required using FreeStyle. No more. Now, you can also apply image sharpening globally via the driver control panel (right-click on desktop > Nvidia Control Panel), complete with adjustable sharpening sliders. Different games look better with different sharpening levels, so Nvidia’s new feature allows you to set per-game sharpening.
GPU upscaling is also offered. This technology works best with games that are rendered below your monitor’s native resolution for enhanced performance, then scaled up to your display’s specifications.
By implementing image sharpening in the Nvidia Control Panel, the company says it should work with all DirectX 9, 11, and 12 games. Vulkan and OpenGL support are “coming soon in a driver update.”
Nvidia’s also boosting support for adaptive sync monitors. First, it’s adding seven more FreeSync monitors to its list of G-Sync Compatible displays, which means they will automatically enable buttery-smooth variable refresh rates when paired with a GeForce graphics card. These are the newly added models:
- Acer CG437K P
- Acer VG272U P
- Acer VG272X
- AOC 27G2G4
- ASUS XG279Q
- Dell AW2720HF
- Lenovo Y27Q-20
Finally, Nvidia’s own hardware-infused G-Sync monitors will now behave better when you’re gaming at screaming-fast speeds over your monitor’s native refresh rate, in an effort to counter-punch AMD’s Radeon Anti-Lag. The company’s Nvidia Ultra Low-Latency technology, which queues frames just-in-time for the GPU, now supports G-Sync. You activate it by setting the Low Latency Mode option to Ultra.
Enabling NULL with V-Sync off improves latency but sacrifices visual fidelity—you’re subjected to screen tearing due to the way the technology works. Conversely, enabling V-Sync to reduce tearing without G-Sync active greatly increases latency. Now, however, you can enable NULL, V-Sync, and G-Sync simultaneously to eliminate screen tearing and get latency results on a par with native game performance with all those technologies turned off. Impressive! Keeping V-Sync and NULL active with G-Sync disabled is still faster, but you’ll still get that nasty screen tearing.
We haven’t had time to play around with the new feature yet, but it sure sounds cool on paper. Near-native latency without ugly screen tearing? Yes please.
The new GeForce drivers and the new GeForce GTX 1660 Super graphics cards should both be available Tuesday.