Reader, I spent two days in the benchmark pit and told me what you’re already guessing. The GeForceRTX5090 is extremely fast and too expensive, with more AI technology than Philip K. Dick’s cheese dream. At least two of these points are sure to run the average graphics card shopper, especially at a time when even game developers suspect generative AI and many of its sambara-like robotic voice nostrils.
However, there’s not much to do about the RTX 5090 cost. at least £1939/$1999, more than hundreds of more than the infamous Spany RTX 4090, the suite of artificial intelligence tools that is purely focused on performance, dare I say – pretty clean. These range from Multi Frame Generation (MFG), which is essentially a DLSS 3-frame GEN but up to twice as fast, to the general upscaling enhancement of DLSS 4, as well as the ability to apply newer DLSS versions to older games. All of this also happens to the rest of the RTX 50 series, slowly falling apart to the Enter RTX range. Therefore, the RTX 5090 is best understood not as a practical GPU purchase per se, but as a Picadily Square Filling Advert for its more affordable siblings.
Read Also:
There’s definitely a sense that Nvidia is going to “Sod It” with this, cutting off the sensitivity brake and generating something that even Hyperen eager might find a bit crazy. In addition to the price, this is the biggest PSU-Muncher to wear a GeForce badge. The Founders Edition I tested is rated to 1000W power and hoover up to 575W at a time. Credits of a redesigned dual-fan cooler to maintain the lid at temperature. Still, I measured the peaks at 74°C, which is very manageable.
That memo: If there is something beautiful about the RTX 5090 None The AI warning is industrial design. This may actually be acknowledged that it’s not hiber – the best looking graphics card I’ve seen in my life, a modern yet mature, precise cut aluminum slice. And unlike the RTX 4090 FE, it’s not as big as it could have come from a time when computers needed their own cupboards. To be precise, it’s a solid 2cm thinner, and ultimately arrests and reverses the trend of obese premium GPUs.
Nvidia GeForce RTX 5090 Review: 4K Benchmark
Ultra HD is a spiritual home for the RTX 5090, just like the RTX 4090 and RTX 3090 before. I had all three in hand, so I introduced them to a new test rig piloted with an Intel Arc B580.
When it comes to unchanged native lesbian gaming performance, the RTX 5090 is definitely the fastest card in all of them new, let alone frame Gen Gubbins. This is the first time I’ve destroyed 100fps on a Cyberpunk 2077 with these settings, and at least a 144Hz monitor (whether it’s worth £2K if it’s not a relatively big question, you can see visible improvements in other tough games like the RTX 4090, like Total War: Warhammer III and F1 24.
Nevertheless, the jump from the RTX 4090 to the RTX 5090 is as large as the former improvement over the RTX 3090, making it difficult to ignore the inconsistent method. For example, the quality level of Forbidden West achieved a maximum of 137fps on the RTX 5090 and 125fps on the RTX 4090.
Meanwhile, the Blackwell Architecture on the RTX 5090 looks a little better at handling Ray Tracing’s performance hits. This was already one of Nvidia’s powerful suits. After enabling the ultra-quality RT effect, Metro Exodus dropped from 163fps to 117fps on the RTX 5090 and from 135fps to 87fps on the RTX 4090.
Next is DLSS 4 and its most numbered component, multi-frame generation. This works similar to frame generation introduced by DLSS 3 in the 40 series. The card retrieves data from the frame the GPU is rendering, generates a frame like a look that can be glided between the rendered ones using AI, exchanges the overall framer rate and exchanges it for an additional touch of input lag. However, the DLSS 3 version creates up to one AI frame for each rendered frame, while the MSG generates three or four, resulting in a scene like this:
It’s certainly an RTX 5090 running Alan Wake II with a fully pass-traced Cyberpunk 2077 and Alan Wake II in 4K – as far as your eyes are concerned – 200fps. Importantly, 4x MFG does not significantly increase latency for what the classic 2x frame generation has already done, nor does it get worse in terms of image quality.
That’s not to say it’s Immaculate Magic. Input lag is input lag, and MFG does not significantly exacerbate this tradeoff, but it does not improve either. Cyberpunk 2077 can be very blurry when making quick camera movements. This was not the AW2 and the Dragon Era: Belguard. Regardless of the number of mimic frames an AI generator can add, it is important to remember how these games will feel in your hands. 218FPS only has snapiness for 64fps gaming mouse. This is because these frames are the only frames that are delivered via a regular PC pipeline that can accurately reflect the input.
With that in mind, visual smoothness teeth Stark Upgrade, and if you’re not playing anything that needs to be targeted for Hyper Axela’s cramps… why not? Frame generation is indeed a luxury in that case, especially if the underlying GPU is strong enough to comfortably get “real” framing benefits from maximum quality settings like pass traces. It’s not a crutch that games without the right low-end settings can be used to boost numbers.
MFG is also part of DLSS 4, both of this new Nvidia generation, covering a variety of tricks that are not exclusive. Therefore, the RTX 5090 is also obligated to showcase technology that will benefit GPUs, dating back to the RTX 20 series, such as the new “trans” model of DLSS 4 and support for the “DLSS Overrides” of the NVIDIA app.
First, the trans model is a rigid method of how DLSS applies best-in-class anti-aliasing. It’s slightly slower than the existing “convolutional neural network” model, averageing 120fps at Cyberpunk 2077 at 4K/ultra versus the older version of 132fps. But it looks sharper and more detailed. Many Nvidia promotional comparison images focus on textures, but what I found most noticeable was the improvements to fine objects like grass and wire fences. While DLSS already looked better than their rivals FSR and Xess, the transformer model cleans them up and sharpens them, bringing the big picture closer to true native lesbian quality than ever before. Something good.
Second, DLSS overrides can also be a huge problem for games that never updated support and lacked it in the first place, with newer, better versions of DLSS. The NVIDIA app still requires some support. You won’t be able to slap DLSS 4 with anything created so far, but there are 75 compatible batches of games at launch, and the options are impressively flexible. For example, if it is not officially implemented, you can force the game to use DLAA anti-aliasing, or get a game with convolutional neural network DLSS support and upgrade to a new trance model. Also, if you have a 50 series GPU like this, you can create games using the DLSS 3 Frame Gen feature. DragonAge: Like The VeilGuard, you can make 4X MFG work completely with 4x MFG.
I haven’t yet seen the value yet, including much of what Nvidia is doing with AI, including their attempts to create creepy, artless NPCs. But it’s much easier to ride on these DLSS advancements.
Nvidia GeForce RTX 5090 Review: 1440p Benchmark
However, it’s time to remind you once more of the RTX 5090 limits before becoming Lovey-Dovey. Buying a top shelf GPU and hooking to reduce the 1440p 1440p screen may seem like a Galaxy Brain setup building. In fact, your mismatched rigmobile will just be sent to crash into a CPU bottleneck.
Again, the RTX 3090 is quite behind, but at this resolution, the RTX 5090 is as powerful as the GPU it replaces. You can open up the gap again by placing it on top with raytracing, but still, it overwhelms the Cyberpunk 2077 by averaging 124fps on the RTX 5090 and 111fps on the RTX 4090, using the PsychoRT Effects and Quality DLSS. It’s not a big deal. MFG is not very useful here either. With many frames already being pushed into the eyes, the normal 2x frame generation will push support games beyond points that rarely perceive more load.
No, the 4K has the RTX 5090. But then, when you try to become practical with a card like this, it feels like the point is missing. Yes, once had the GTX 1080 TI and RTX 2080 Ti, a GPU, which can surpass the range, but is a somewhat achievable high ender. I miss those days too. But isn’t the RTX 5090 an avid device anyway? Is this a concept model that just happened to escape the design wing and sneaks into a production factory?
Let’s be authentic: you weren’t going to buy this. I wasn’t going to buy this. Even people Maybe Perhaps when they’re not busy running the general, they probably have at least one eye on the RTX 5080 too. It’s way beyond our standards, so it doesn’t seem like much to hate it either. I don’t care about the GPU itself, and Nvidia certainly doesn’t. They are probably already planning the RTX 6090 for £2,500.
Instead, let’s take a look at this as to what it actually is. The eye-catching exercises with the DLSS feature are much more likely to create waves than a single graphics card. Even if you haven’t got my money, you have me there, RTX 5090.
This review is based on retail units provided by the manufacturer.