David Bridges
The $2,000 Nvidia RTX 5090 graphics card offers such immense power that it sparked a question in my mind: is there a limit to the number of frames in PC gaming? This thought emerged while I was immersed in Dragon Age: The Veilguard, utilizing Nvidia’s impressive Multi Frame Generation feature introduced with the DLSS 4 update. During an intense battle sequence, the frame rate soared beyond 360 FPS at the highest graphical settings. Even in moments of relative calm, the frame rate remained astonishingly high. While the new Blackwell architecture plays a significant role, the graphics card serves merely as a vehicle for advanced AI technology and the generation of “fake frames.”
At the heart of this innovation is the Multi Frame Generation technology. With Nvidia’s recent update to its AI upscaler, DLSS 4, the latest Nvidia GPUs can create up to three additional frames between two rendered frames. When paired with a powerful graphics card, this multi-frame capability is akin to adding a luxurious cherry on top of an extravagant gold-leaf sundae. But what does it truly mean to experience a single-player RPG at frame rates of 120 FPS compared to an astonishing 360 FPS? After extensively playing several major titles that harness the new DLSS 4 features, I found myself questioning the necessity of such high frame rates, aside from the simple thrill of watching the numbers climb higher.
This review does not aim to be an exhaustive evaluation of Nvidia’s new GPU. I lack direct access to the $1,600 RTX 4090, which would allow for an adequate comparison with the RTX 5090 Founders Edition. However, I can provide insights by comparing it to the RTX 4080 Super and RTX 4070 Ti Super, both of which are excellent cards priced significantly lower than the 5090. Regardless of the high-end specifications, high-performance graphics cards like the 5090 ultimately depend on software capabilities. For instance, if you aim to play Cyberpunk 2077 at 4K resolution with maximum settings and ray tracing enabled, you won’t achieve 60 FPS without DLSS. However, enabling DLSS at balanced settings can push the frame rate above 100 FPS. While frame generation provides an exhilarating numerical experience, utilizing AI-generated frames does come with its share of unintended consequences.
Mark your calendars: the Nvidia GeForce RTX 5090 Founders Edition is set to officially launch on January 30.
Experience High-Performance Gaming: How Does the RTX 5090 Enhance Gameplay?
The specifications of the Nvidia GeForce RTX 5090 are as impressive as its price tag suggests. Built on the cutting-edge Blackwell architecture, the Founders Edition features a dual-fan configuration, boasting a staggering 32 GB of VRAM, 21,760 CUDA cores, a base clock speed of 2.01 GHz, and a boost clock speed reaching 2.41 GHz. Additionally, it is equipped with advanced 4th-generation ray tracing cores and promises an exceptional performance level of 318 TFLOPS. While these numbers are remarkable, Nvidia has primarily focused its efforts on enhancing the 5th-generation Tensor cores, which deliver an astounding 3,352 TOPS of AI performance.
The role of AI is crucial in this context. Justifying the steep price of the RTX 5090 would be challenging without the advancements brought by the Blackwell architecture and the innovative multi-frame generation capabilities. Similar to the frame generation introduced with DLSS 3, this technology inserts a generated frame between two rendered frames, but Nvidia claims that Multi Frame Generation has improved significantly. It is now more efficient and operates 40% faster, thanks to the inclusion of an AI Management Processor on the GPU itself that effectively manages various AI tasks.
To enhance frame generation, the Blackwell architecture employs a technique known as “flip metering.” This system strategically slots generated frames to minimize latency. Even with these advancements, Nvidia encourages users to utilize RTX Reflex 2 to further reduce latency. If all this sounds like a lot to add to a new graphics processor’s traditional capabilities, it certainly is.
It’s worth noting that some games may take time to update their user interfaces to fully support Multi Frame Generation or other new DLSS 4 features. Fortunately, the Nvidia application includes a DLSS override feature, allowing users to force an upgrade to the latest upscaler and frame generation technologies. Users can select from 4x, 3x, or 2x frame generation, depending on their preferences.
While older RTX cards will receive some updates to DLSS 4, the RTX 40-series will only benefit from enhancements to the transformer model DLSS and limited access to 2x frame generation. Nvidia has indicated that the Blackwell architecture is essential for utilizing Multi Frame Generation effectively.
Exploring Gameplay Dynamics: How Does Frame Generation Impact Your Experience?
For my testing, I utilized Nvidia’s flagship RTX 5090 in a prebuilt Origin PC Neuron 3500X, which came equipped with an Intel Core Ultra 285K CPU and 32 GB of LDDR5X RAM. The system also featured a 1000 W 80 Plus Gold power supply, which Nvidia recommends as the minimum for the 5090. Priced at nearly $3,400, this desktop tower represents a significant investment. Although a 1000 W power supply is standard for most high-end gaming PCs, it was previously considered essential for providing ample headroom and future-proofing. Given that the RTX 5090 itself costs $2,000, the necessity of additional high-quality components further elevates the cost of building a complete high-performance gaming rig.
Upon launching a game, witnessing the frame counter soar can be both thrilling and unsettling. There’s an instinctive urge to verify that everything is functioning correctly, and in my experience, I struggled to identify any significant visual discrepancies. While examining the foliage in Dragon Age: The Veilguard, I found no noticeable issues with sharpness or texture quality.
I also tested the card in other demanding titles like Hogwarts Legacy and Cyberpunk 2077. The thought of experiencing Cyberpunk at 200 FPS seemed unfathomable, yet here I was witnessing it firsthand. The Multi Frame Generation feature enabled me to run path tracing at ultra settings while maintaining high FPS, leading any onlookers to believe I was manipulating the system (which, in a way, I was).
While exploring the vibrant streets of Night City on a bike, I did not encounter any significant stuttering. However, I did notice some flickering lights on Jackie’s bike while speeding around. This flickering was evident at all levels of frame generation, although it was most pronounced when using the highest frame generation settings. You can observe this anomaly in the video linked above.
Despite being somewhat distracted by the UI flickering, the gameplay remained fully enjoyable. I did not experience any floatiness in controls or visual artifacts, though I can’t rule out the presence of other subtle issues that I may not have detected. With the implementation of Reflex 2, I also found no responsiveness problems during gameplay.
With Frame Generation disabled, Cyberpunk ran at around 100 FPS with DLSS set to balanced mode. Without the DLSS enhancement, the frame rate dropped below 60 FPS. The RTX 5090 can handle the game quite well without frame generation, and while the extra frames are impressive, they may not necessarily enhance the overall experience. Some players might find these “fake frames” to be more distracting than beneficial.
I was genuinely surprised by how normal games appeared while operating at these incredibly high frame rates. In Dragon Age: The Veilguard, one cutscene seemed to play too quickly, but this issue was not repeated in subsequent instances. In Alan Wake II, I struggled to exceed 60 FPS with path tracing on Ultra and all other settings maxed out, even with the RTX 5090 and DLSS enabled in balanced mode. However, when activating 4x Multi Frame Generation, the frame rate soared past 190 FPS. While I did encounter some minor pop-in issues, this is a well-documented challenge when running the game at maximum settings on PC. Upon testing, I discovered that the stark difference between 70 FPS without frame generation and 250 FPS with 4x frame generation did not negatively impact gameplay.
Evaluating the Necessity of the RTX 5090’s Frame Generation Technology
But does the RTX 5090’s frame generation truly enhance gameplay? I typically test games at 4K resolution on an AOC U27G3X monitor, which offers a maximum refresh rate of 120 Hz. This refresh rate is adequate for most GPUs, including the 5090, unless you’re capitalizing on multi-frame generation. Without utilizing 2x, 3x, or 4x frame generation, I would only anticipate achieving around 100 FPS using medium DLSS upscaling without path tracing in the most demanding games.
To fully realize the potential of 360 FPS, I would need an exceptionally expensive monitor, such as the LG Ultragear, capable of reaching 480 Hz refresh rates. However, this may come at the expense of maintaining 4K resolution. The most recent Consumer Electronics Show (CES) was flooded with stunning 240 Hz 4K OLED monitors from a variety of manufacturers. While it’s possible to acquire a gorgeous, high-end curved monitor that delivers both 4K resolution and 240 Hz for over $1,000, finding one that can handle the sheer number of frames generated can be quite a challenge. Even LG’s acclaimed bendable 5K2K monitor only supports a maximum of 165 Hz at higher resolutions.
While many monitors can achieve 240 Hz, very few can support higher refresh rates. This limitation stems from the fact that there comes a point where an increased frame rate ceases to make a significant impact on the gaming experience. Even professional esports competitors often utilize 240 Hz monitors, as casual gamers may not perceive a substantial difference between 120 FPS and 240 FPS. For example, I tested the RTX 5090 with Marvel Rivals, a game that doesn’t place excessive demands on a gaming PC. The game achieved over 300 FPS with multi-frame generation enabled. However, as a casual gamer, I cannot definitively state whether the generated frames were beneficial or detrimental to my performance. In either case, they did not improve my aim.
Nvidia’s CEO, Jensen Huang, acknowledged in a recent press conference that the $2,000 price tag for the RTX 5090 reflects its positioning for a niche market; it is designed for gamers who desire nothing less than the absolute best, regardless of cost. This GPU caters to enthusiasts who are willing to invest heavily to achieve top-tier performance. Owning the best graphics card necessitates accompanying components, including high-end CPUs, top-rated power supplies, and some of the priciest monitors available. This card is crafted for individuals who view cost as no object. At that point, one must question the purpose of a review; the 5090’s superiority over the 4090 is already evident, especially considering it is priced $400 higher than Nvidia’s previous flagship model.
To make frame generation truly useful, a game must consistently run at a stable 60 FPS. This is why I am more intrigued by the upcoming RTX 5070 and 5070 Ti, as they potentially serve gamers who cannot afford to spend their entire monthly rent or mortgage payment on a single graphics card. For these budget-conscious gamers, Multi Frame Generation could represent a more cost-effective solution. This is why I am particularly excited about the yet-to-be-announced RTX 5060; its performance will depend on how effectively it can maintain standard frame rates with fully rendered frames. Budget gamers are accustomed to making compromises, while those who invest $2,000 in a graphics card expect nothing less than perfection.
