
Gamers invest significant effort and even more capital into attempting to elevate their FPS counter progressively. Although Nvidia is not the only GPU manufacturer asking users to spend additional money for increased frames, it might be pioneering the inquiry: Will you accept when—not just some, but the majority of your frames are produced through AI rather than rendered? We’ve observed numerous demonstrations of Nvidia’s multi-frame gen in real life. Purely judging by the FPS statistics, it stands to be a pivotal point for PC gaming in recent years, provided you have the resources to purchase another graphics card.
In a private media briefing with journalists, Nvidia emphasized the strength of its new, hefty, and pricey RTX 50-series cards. This collection is spearheaded by the $2,000 RTX 5090, featuring a desktop GPU armed with 32GB of VRAM. Naturally, it’s a stronger card. It comprises 21,760 CUDA cores along with the new Blackwell shader cores. Nonetheless, the VRAM was explicitly enhanced to aid AI processing. Aside from its graphical prowess, Nvidia’s GPU announcement heavily revolved around its 5th-gen Tensor cores and the remarkable 3,352 TOPS of generative AI processing power.
The RTX 5090 is a GPU aimed towards both AI creators and gaming enthusiasts. But why does that matter? Does AI processing have any value for gamers? Indeed, at least in terms of acquiring those additional frames. Nvidia highlighted its multi-frame generation skills, yet the narrative is broader. The advancement to its upscaler technology, DLSS 4, serves as the cornerstone of Nvidia’s aspirations for its latest cards. Nvidia asserts a game like Star Wars: Outlaws will operate at merely 50 FPS at 4K with max settings on a 5090 without upscaling. Incorporating DLSS 4, along with all additional features, Nvidia claims you can exceed 250 FPS with reduced latency compared to normal. It sounds almost too wonderful to be true, and for purists in PC gaming, it possibly is.
In What Way Does Multi-Frame Gen Enhance Game Performance?

Essentially, Nvidia’s frame gen employs a generating model to anticipate the following frame in a sequence, subsequently inserting a new frame based on the standard, rendered in-game image. The existing frame gen on DLSS 3.5 is supported by select titles to generate frames on a 1-to-1 basis. Earlier iterations of single frame gen could offer a marked boost in framerates, whereas a refreshed version should elevate framerates to a greater extent. Nvidia declares its new frame generation model is 40% swifter than before and utilizes less VRAM; additionally, it can model multiple frames based on a single rendered scene and place it before the subsequent rendered frame without affecting latency.
Nvidia asserted that if a game executes all of DLSS 4, the GPU commands five AI models simultaneously, hence requiring the 50-series cards equipped with their new Tensor Cores to operate multi-frame gen. Observed firsthand, the framerates are instantaneously remarkable. A system with a 5090 running Cyberpunk 2077 at 90 FPS might approach 170 FPS in certain scenes with multi-frame gen. An RTX 4090 PC achieving 130 FPS in Alan Wake 2 with ray reconstruction could be nearing 300 FPS in the identical scene with a 5090 and multi-frame gen.
The technology is already awe-inspiring, yet the true consideration for gamers is determining whether they can endure the notion of viewing “synthetic frames,” as critics describe them, rather than genuinely rendered frames. In a particular demo, a spectator noted a sudden flicker from a light source, though this might have been an issue related to the monitor. Concerning detail, I struggled to discern a noticeable distinction between a game with generated frames and one without. The advantage is that Nvidia’s enhanced ray reconstruction aims to bolster clarity in fine details, like overhead wires or chain-link fences.
For gameplay, Nvidia ensured Reflex 2 was activated to diminish latency. In a game like Black Myth Wukong, there wasn’t any apparent variation in a game operating at 90 FPS versus one running at elevated framerates, aside from the most negligible improvement in fluidity. The discrepancy between 170 FPS and 230 FPS in a game like Marvel Rivals might be more pronounced. Nonetheless, that’s only applicable if you’re an aspiring pro gamer playing a shooter at that caliber and require utmost accuracy.
These aren’t “synthetic frames” but they aren’t generated by the PC’s processors either. Multi-frame gen functions as a sleight of hand necessitating distraction. Users remain immersed in gaming and revel in the framerate ticker, potentially overlooking any visual inconsistencies when these—undoubtedly—materialize. Parsing this takes time, something infeasible even within a few hours of demos. We’ll need to acquire these new cards ourselves to assess the impact on our gaming adventures.
Which Games Receive the Most Advantage from Multi-Frame Gen?

Nvidia projects 75 titles and software that are going to accommodate the multi-frame generation, involving forthcoming releases like Doom: The Dark Ages and Dune: Awakening. Amidst the ones initially compatible, there are some unique entries, such as the climbing title Jusant and Deep Rock Galactic. Each doesn’t put heavy demands on the GPU during execution. The almost decade-old Deep Rock Galactic commonly places more strain on the CPU. A superior card can hike up framerates but not to an extent as in more demanding titles. By appending multi-frame generation, the frame count you’d achieve may likely surpass what most upscale monitors can accommodate within an economical range.
The advanced RTX 40-series cards, particularly the 4090, represented the zenith of what common users could own for desktop visuals. While AMD and now Intel contest the low-to-mid-scale GPU segment, Nvidia solely assures “the best.” CEO Jensen Huang affirmed this during a Q&A, a day succeeding his CES keynote presentation. Nonetheless, the 4090 remains untapped entirely. One can still purchase it and realize remarkable performance when united with an elite CPU.
Running a 5090 asks for no less than 575W of energy, henceforth, you’ll probably require a hefty, certified PSU at the bare minimum. Pairing it with the newest AMD or Intel desktop CPUs would be advisable, although present or forthcoming top-tier processors might pose bottleneck issues. While the 4090 is already plenty, being secure and understood, the 5090 is monumental, demanding a reason to operate it. Still, possessing a costly, elite 240 Hz monitor and desiring to experience games at their apex in both aesthetics and function, the 5090 vows to realize that for a substantial sum.
Which Gamers Will Benefit From the DLSS Updates?

This highlights the significance of multi-frame generation. Solely the innovative 50-series cards will possess this characteristic. The 40-series will feature a restored single-frame generation, yet those GPUs, together with the 30-series and 20-series cards, will gain access to the transformer model super-resolution, DLAA, and updated ray reconstruction. Nvidia’s internal statistics suggest almost 80% of gamers employ DLSS. It’s a robust adoption rate, although Nvidia has not identified which players are utilizing the company’s more affordable or older graphics cards.
No details have been disclosed by Nvidia about the awaited RTX 5060 and RTX 5050. The more budget-oriented cards are rumored to contain 8 GB of VRAM. It’s what also accompanies the 5070, and even though it might seem scant, Nvidia declared that the card can support multi-frame gen. For budget players, frame generation could be more advantageous than those willing and able to spend $2,000 on a fresh GPU. They are less inclined to worry about “fake frames” criticisms, provided that they can enjoy demanding games at feasible framerates.
It’s confirmed that the $550 RTX 5070 can manage multi-frame generation, thus Nvidia marketed it to match the $1,600 RTX 4090’s capability. This is promising for budget-tier desktop PCs and laptops. Ultimately, what matters to users is their acceptance of the possibility that their frames are “fake” or generated artificially. If the goal is to fully appreciate what developers designed, then the origin of frames is irrelevant so long as they enrich the experience.