Saturday, November 26, 2022 | 02:47 pm

NVIDIA Discontinues the GeForce RTX 4080 12GB

NVIDIA Discontinues the GeForce RTX 4080 12GB

Were you confused when NVIDIA introduced two GeForce RTX 4080 models with significantly different capabilities? You’re not alone. NVIDIA is “unlaunching” the 12GB RTX 4080 over concerns they’re confusing. While the company still believes the 12GB version is a solid video card, it said the lower-spec model is “not named right.” The company didn’t say if or how it might relaunch the card, and didn’t elaborate further when reached for comment.

The 16GB RTX 4080 is still on track to launch November 16th at a price of $1,199 for NVIDIA’s Founders Edition. This effectively raises the base price for RTX 40-level hardware. This might not be as bad as it sounds, minds you. The 12GB edition was supposed to start at $899, although its dependence on third-party manufacturers was likely to raise real-world pricing higher.

The muddled messaging largely stems from differences beyond RAM quantities. Where the 12GB model includes 7,680 CUDA (general-purpose processing) cores and a 192-bit memory bus, its higher-end counterpart offers 9,728 cores and a 256-bit bus. You’re getting a considerably slower GPU with the 12GB unit despite the RTX 4080 branding, and the gaps are sometimes huge. NVIDIA’s own benchmarks showed the 12GB board trailing its 16GB sibling by up to 30 percent in well-known games. That could be more than a little disappointing if you bought the 12GB card expecting similar performance outside of RAM-dependent situations.

The cancellation won’t necessarily create much of a headache for NVIDIA. The RTX 4090 launch reportedly created lines at stores, and the 4080 could easily be the go-to GPU for gamers who can’t quite justify the $1,599 flagship. Still, it’s rare to see a misstep like this – NVIDIA misjudged the market to the point where it had to axe a graphics card before it reached customers.

Following months of anticipation and controversy among its add-in board partners, NVIDIA’s 40 series GPUs are finally here. The company unveiled the GeForce RTX 4090 and GeForce RTX 4080 today at its GTC 2022 keynote. Taking full advantage of its new “Ada Lovelace” architecture, NVIDIA says the two GPUs offer significantly better ray tracing performance. The company worked with Taiwan Semiconductor Manufacturing Company (TSMC) to co-develop a new “4N” fabrication process that NVIDIA says is up to two times more power efficient than the 8nm process it used for its 30 Series cards. 

Ray tracing performance is significantly improved thanks to Ada Lovelace featuring NVIDIA’s new third-generation RT Cores, and the inclusion of a new rendering technique called Shader Execution Reordering and DLSS 3.0. In some games, NVIDIA said you can expect two to three times better ray tracing performance than what was possible with its Ampere GPUs. The company demoed Cyberpunk 2077 running at a near-consistent 100 frames per second with all of the game’s ray tracing features set to max. NVIDIA said rasterization performance is up to two times faster thanks to the new architecture.  

The first of NVIDIA’s new Ada Lovelace GPUs will arrive next month when the GeForce RTX 4090 goes on for sale for $1,599 on October 12th. With 24GB of GDDR6X memory, NVIDIA claims its latest flagship is two to four times faster than the 3090 Ti while consuming the same amount of power. Good thing too because it’s starting at $100 more than its predecessor. Inside of the RTX 4090, NVIDIA has managed to fit 16,384 CUDA Cores clocked at a base speed of 2.23GHz.   

Alongside the 4090, NVIDIA will offer two different variants of the RTX 4080. The base model, starting at $899, features 12GB of GDDR6X memory, while the 16GB version will set you back a cool $1,199. Both configurations will arrive sometime in November. However, NVIDIA will only sell a Founders Edition model of the more expensive model. For the 12GB version, you’ll need to look to the company’s partners, which may make it hard to find models that actually start at $899.  

In terms of performance, the 16GB 4080 features 9,728 Cuda Cores and a base clock of 2.21GHz, with a maximum boost clock of 2.51GHz. Meanwhile, the 12GB model features a more modest 7,680 CUDA Cores but 100MHz faster base and boost clocks. Thankfully, you probably won’t need to upgrade your power supply if you plan to upgrade from a 3070 or 3080, with NVIDIA recommending a 700-watt PSU for the 12GB variant and a 750-watt power supply for its more powerful sibling. However, should you decide to buy a new PSU, you’ll want to wait until more ATX 3.0 PSUs arrive later this year. That’s because at least the Founders Edition models will support the new PCIe Gen-5 16-pin connector standard. That said, NVIDIA will also include an 8-pin adapter for those who don’t want to rewire their system.    

NVIDIA’s 40 Series GPUs arrive at a difficult time for the company. For much of the pandemic, it was impossible to buy the latest GeForce GPUs at MSRP due to demand from both gamers and crypto miners. That all changed in recent months due to the recent crypto crash and Ethereum’s much-anticipated switch to proof-of-stake minting. As a result of those events, the used market was flooded with 30 series GPUs, making it nearly impossible for the company’s AIB partners and retailers to sell new video cards at MSRP.

Surprise! The iPhone 14 is pretty repairable, it turns out. This week, Cherlynn and Devindra chat with Engadget’s Sam Rutherford about this move towards greater repairability and what it means for future iPhones. Also, they dive into NVIDIA’s powerful (and expensive!) new RTX 4080 and 4090 GPUs. Sure, they’re faster than before, but does anyone really need all that power?

Listen above, or subscribe on your podcast app of choice. If you’ve got suggestions or topics you’d like covered on the show, be sure to email us or drop a note in the comments! And be sure to check out our other podcasts, the Morning After and Engadget News!

Portal 3 may never happen, but at least we’ve got a new way to experience the original teleporting puzzle shooter. Today during his GTC keynote, NVIDIA CEO Jensen Huang announced Portal with RTX, a mod that adds support for real-time ray tracing and DLSS 3. Judging from the short trailer, it looks like the Portal we all know and love, except now the lighting around portals bleeds into their surroundings, and just about every surface is deliciously reflective. Similar to what we saw with Minecraft RTX, Portal’s ray tracing mod adds a tremendous amount of depth to a very familiar game. And thanks to DLSS 3, the latest version of NVIDIA’s supersampling technology, it also performs smoothly with plenty of RTX bells and whistles turned on. This footage likely came from the obscenely powerful RTX 4090, but it’ll be interesting to see how well Portal with RTX performs on NVIDIA’s older 2000-series cards. Current Portal owners will be able to play the RTX mod in November.  

Huang says the company developed the RTX mod inside of its Omniverse environment. To take that concept further, NVIDIA is also launching RTX Remix, an application that will let you capture existing game scenes and tweak their objects and environments with high-resolution textures and realistic lighting. The company’s AI tools can automatically give materials “physically accurate” properties—a ceiling in Morrowind, for example, becomes reflective after going through RTX Remix. You’ll be able to export remixed scenes as mods, and other players will be able to play them through the RTX renderer. 

NVIDIA’s GeForce RTX 40 series GPUs won’t just rely on brute force to deliver high-performance visuals. The company has unveiled Deep Learning Super Sampling 3 (aka DLSS 3), a new version of its AI-based rendering accelerator. Rather than generating ‘only’ pixels, the third-gen technology can create entire new frames independently. It’s a bit like the frame interpolation you see (and sometimes despise) with TVs, although this is clearly more sophisticated – NVIDIA is improving performance, not just smoothing out the video.

The technique relies on both fourth-gen Tensor Cores and an “Optical Flow Accelerator” that predicts movement in a scene by comparing two high-resolution frames and generating intermediate frames. As it doesn’t involve a computer’s main processor, the approach is particularly helpful for Microsoft Flight Simulator and other games that are typically CPU-limited. A new detail setting in Cyberpunk 2077 runs at 62FPS in 4K resolution using DLSS2 in NVIDIA’s tests but jumps beyond 100FPS with DLSS 3.

Roughly 35 apps and games will offer DLSS 3 support early on. This includes Portal RTX, older titles like The Witcher 3: Wild Hunt, and releases based on Unreal Engine 4 and 5.

It’s too soon to say how well DLSS 3 works in practice. NVIDIA is choosing games that make the most of DLSS, and the technology might not help as much with less constrained titles. Nonetheless, this might be useful for ensuring that more of your games are consistently smooth. Provided, of course, that you’re willing to spend the $899-plus GPU makers are currently asking for RTX 40-based video cards.

How do you go about reviewing something like NVIDIA’s RTX 4090? Just looking at its specs alone, it’s obviously the fastest consumer GPU we’ve ever seen. So sure, I can tell you that I can play just about anything in 4K with ray tracing and every graphical nicety turned on. Hell, it can even scale up to 8K if you’re a masochist. For a $1,599 video card, it is damn well better. But the real question is, who is this thing actually for?

Benching the RTX 4090 against NVIDIA and AMD’s older hardware is practically pointless. Of course, it’s far faster. Of course, it’ll make you jealous. If you’ve got the cash and you’re itching to upgrade, go with God (or NVIDIA’s leather-clad CEO Jensen Huang, as the company’s fans see him). But for anyone else who doesn’t need bleeding-edge hardware, it exists purely as an object of lust. Sure, you could wait for the upcoming RTX 4080 cards, or whatever AMD has in the works, but it’s not a 4090. Just like the last generation of GPUs, NVIDIA is throwing down the gauntlet with a power-hungry card for the most hardcore gamers and creators.

If your mind isn’t made up, I assume you’re here just to see how much of a beast the 4090 is. And let me tell you, it’s a stunning thing to behold. Weighing in at 4.8 pounds, and approaching the size of the PlayStation 5, the RTX 4090 is a triple-slot GPU that will dominate whatever case it’s in. Seriously, if you’re thinking of getting it, be sure to measure your PC to ensure you can fit a nearly foot-long card that’s close to 5 inches thick.

Be prepared to upgrade your power supply too: The 4090 has a high 450W TDP (the same thermal design profile as the 3090 Ti) and it requires an 850W PSU. (Some third-party companies are pushing that demand to 1200W PSUs!) While it can be powered by a single PCIe 5.0 cable, there still aren’t many of those PSUs on the market, so most people will likely end up using four 8-pin adapters. I cursed Jensen’s name when I realized I needed to string another PSU line, after tidying up all of my cables.

Beyond its obscene power demands, though, NVIDIA hasn’t changed much about the 4090 Founder Edition’s design from its previous model: It’s still a high-end, all-metal card with a massive vapor chamber, heatsink array, and two fans on opposite sides. NVIDIA claims they can push 20 percent more air than the 3090 Ti – in my testing, that meant the 4090 stayed at a relatively cool 70C under load.

What’s truly special about the RTX 4090, though, is everything under the hood. It features the company’s new “Ada Lovelace” architecture (named after the world’s first computer programmer, though I wonder if NVIDIA pays any royalties to turn her name into marketing). It has 16,384 CUDA cores (almost 6,000 more than the 3090 Ti), a base clock speed of 2.23GHz (boost to 2.52GHz), and 24GB of GDDR6X RAM. With figures like these, the upcoming RTX 4080 cards (which start with 7,680 CUDA cores) seem puny in comparison.

And really, that seems like the point of dropping the 4090 before the rest of NVIDIA’s new GPUs. It’s like a heavenly body so massive it warps space-time around it. This is the new standard. What other GPU can get you 135fps in Cyberpunk 2049 while playing in 4K with maxed-out graphics and ray tracing? 

To be clear, though, the 4090 isn’t just about brute-force power. It was able to reach that killer Cyberpunk framerate by using DLSS 3, NVIDIA’s latest AI upscaling technology that can now generate entire frames of imagery on its own. (That’s in addition to upscaling lower-resolution textures using AI, like earlier versions.) DLSS 3 helped A Plague Tale Requiem perform more than twice as fast while running in 4K, delivering around 175fps (up from 74fps).

The RTX 4090 had no trouble delivering 107 fps in Control while playing in 4K with high ray tracing settings. But you know what’s even better? Getting a solid 128 fps when I flipped on an older version of DLSS. It’s just unfortunate it doesn’t support DLSS 3 yet, because I’m sure it would eke out even better performance. Even though the game was actually being rendered in 1,440p, to my eye, DLSS still does a stunning job of making that seem like 4K. (I tested the 4090 alongside Samsung’s 55-inch Arc monitor, giving me a much larger view than my typical 34-inch ultrawide screen. If there were graphical anomalies, I would have seen them.) 

I was particularly interested in stressing ray tracing performance on the 4090 because that was a feature that still managed to bring NVIDIA’s 30-series cards to their knees. It enables more realistic lighting, shadows, and reflections. For most, I’d wager the graphical facelift it delivers would be more impressive than a skyrocketing framerate count. So it’s a wonder to see an NVIDIA card that can finally deliver 4K and solid ray tracing beyond 100fps. Is that worth $1,599, though? That remains unclear, especially since we don’t know how the rest of the 40-series cards will compete.

If you’re looking for a video card that can do more than just a game, the 4090 may make more sense. In the Blender 3D rendering benchmark, it scored twice as high as the RTX 3090 Ti, a GPU released earlier this year for an eye-watering $1,999. (Let’s have a moment of silence for the poor souls who jumped on that card.) When it came to transcoding a short 4K clip into 1080p, the RTX 4090 was also 10 seconds faster than the 3080 Ti. That could certainly add up if you’re rendering longer clips, episodes, or feature films.

It’s hard not to covet the RTX 4090, especially once you see what it’s capable of. It’s a glimpse into a world where we can finally get uncompromised ray tracing. But with the $899 and $1,199 RTX 4080 cards on the horizon, it’s tough to drop the price of an entire computer just to get the best frame rates in town There’s just so much more to consider these days. You could pair up one of those 4080s with a Steam Deck, for example, and bring the joys of PC gaming on the road and all over your home. Sure, you won’t have the prestige of being in the 4090 clubs, but you’ll probably end up having more fun.