Nvidia GeForce RTX 3080: News, rumors, and everything we know so far

Nearly two years after bringing ray tracing to the mainstream with its Turing cards, Nvidia has announced its second-generation RTX graphics cards, including the RTX 3090, 3080, and 3070. They’re all based on the new Ampere microarchitecture, and Nvidia claims they boast a monumental leap in performance over their predecessors.

The flagship among the three is the RTX 3080. Here’s everything we know about it so far.

Release date and price

The RTX 3080 was announced on September 1, and it will begin shipping on September 17.

The price will remain at $699, which is the same price as the RTX 2080 and 2080 Super. The larger and more powerful card of the generation, the RTX 3090, will cost $1,500. The 3090, in some sense, replaces both the RTX 2080 Ti and RTX Titan. It will be sold in prebuilt gaming PCs, though it features 24GB of GDDR6X memory for workstation performance in fields like data science and media production.

The RTX 3090 will be available starting on September 24.

The cheapest of the three is the RTX 3070, which replaces the 2070 Super and costs $499. The RTX 3070 will be available sometime in October.

Design

The RTX 3080 Founders Edition features a new “unibody” design that’s a bit more modest compared to the reflective aluminum of the RTX 2000-series cards. We found the card to be quite classy in our review, with an angular design that features a matte, dark grey finish with white backlighting instead of the traditional green hue used on prior GeForce cards.

Early leaked images revealed a very industrial-looking card with a dual-fan design, and they turned out to be accurate. Marking a stark departure from older graphics cards where all the fans are aligned on the same side, the RTX 3080 will have a fan on each side of the card in an effort to better control the flow of hot and cold air throughout the desktop. Nvidia calls its thermal solution a “dual axial flow,” and says it results in twice the cooling performance of its predecessor.

The RTX 3090 uses this same design, only it’s much larger, as the card utilizes a triple-slot format. The RTX 3070, meanwhile, uses a more traditional dual-fan system. For a size comparison, check out the photo below.

The rumor mill has been spinning over a new 12-pin PCIe power interface system for some of these new cards, and it appears that the RTX 3080 and RTX 3090 will make use of the low-profile connectors. The company stated that the new connector combines the power of dual 8-pin connectors in a space-saving design. The RTX 3070, meanwhile, uses just a single eight-pin connector.

Port selection includes three HDMI 2.1 ports and a DisplayPort 1.3a. Importantly, HDMI 2.1 supports a variable fresh rate of up to 120Hz in 4K. New OLED TVs have begun to support this new standard, which could open up more options for PC gaming in the living room.

Specs

Nvidia made some ambitious claims about its new RTX 3000-series cards. First off, it said that both the RTX 3080 and RTX 3070 are more powerful than the current RTX 2080 Ti. As you can see from the chart below, it’s not hard to see how that might be true.

CUDA cores Memory Memory interface Boost clock Graphics card power
RTX 3090 10496 24GB GDDR6X 384-bit 1.70GHz 350w
RTX 3080 8704 10GB GDDR6X 320-bit 1.71GHz 320w
RTX 3070 5888 8GB GDDR6 256-bit 1.73GHz 220w
RTX 2080 Ti 4352 11GB GDDR6 352-bit 1.54GHz 250w
RTX 2080 Super 3072 8GB GDDR6 256-bit 1.82GHz 250w
RTX 2070 Super 2560 8GB DRR6 256-bit 1.77GHz 215w

The RTX 3080 features 8,704 CUDA cores, making for a 65% leap over the RTX 2080 Super and a 50% increase over the RTX 2080 Ti. It’s also a more power-hungry card rated at 320 watts versus the standard 250 watts of the RTX 2080 Ti and RTX 2080 Super.

The RTX 3080 also features 10GB of GDDR6X memory at 19Gbps, which is as fast as it gets. Nvidia said it had collaborated with Micron for the G6X video RAM on the card, which according to the memory partner, delivers a bandwidth of up to 1TB per second. The new Ampere RTX 3080 GPU comes with 30 shader TFLOPs, up from 11; 58 ray tracing TFLOPs, up from 34; and 238 tensor TFLOPs, compared to 89.

Nvidia is also bundling in its new RTX I/O to help improve game loading times on its GPUs. There are three components in play here for the new RTX I/O, including new APIs for fast loading and streaming directly from SSD to GPU memory, GPU lossless decompression, and collaboration with Microsoft for direct storage for Windows.

The result of all this, Nvidia claims, is twice the performance of the original RTX 2080. In real-life performance, Nvidia says the RTX 3080 can consistently run games at 60 fps (frames per second) in 4K with RTX turned on, and though performance varies widely across different titles, we found in our review that 4K at 60 fps stays. Even ray tracing in 4K at decent frame rates is possible if you turn on DLSS.

Even bigger claims were made about the RTX 3090, which can reportedly play games in 8K at 60 fps. This is done using DLSS 2.0, of course, rather than native rendering. The RTX 3070, meanwhile, is supposedly 1.6 times faster than the RTX 2070.

Performance

The new cards are all built on the new Ampere microarchitecture, which is based on Samsung’s 8nm processor, featuring 28 billion transistors. Ampere uses second-generation RT cores and third-generation Tensor cores to boost performance.

Nvidia says performance and energy efficiency is double that of Turing, and works with DLSS 2.0 to denoise as scenes are rendered in real time with cinematic quality. Nvidia claims that the more ray tracing is done, the greater the performance.

“You can see dramatic visual quality jump of Ampere,” Nvidia CEO Jensen Huang said, noting that the RTX 3080 is twice as fast as the Turing-based RTX 2080.

In a subsequent demo uploaded to YouTube, Nvidia showcased the performance of its Ampere-based RTX 3080. The game capture showed that the GPU can handle Doom Eternal at 4K resolution and maintain approximately 100 fps performance rather smoothly.

While Nvidia claimed that the RTX 3080 delivers twice the performance of the RTX 2080 that it will replace, the company did not reveal which benchmarks or tools it had used in making the comparisons.

In our review of the card, we found the performance of the RTX 3080 to be quite impressive. In our 3DMark Time Spy test, the performance of the RTX 3090 was 15% better than the RTX 2080 Ti, despite the latter costing $500 more than Nvidia’s newest offering, and 28% better than the RTX 2080. Compared to the 100% performance uplift quoted by Nvidia, the difference here could be a result of the change in architecture moving from Turing to Ampere.

Though Ampere boasts 50% more CUDA cores, the way each core works is also different on Nvidia’s latest platform.

Prior to the RTX 3000 series unveiling, one leaked internal benchmark showed the performance of the RTX 3090 versus the RTX 2080 Ti in some popular games, showing a 2x improvement in a few titles.

Image source: Twitter user @yuten0x

While these improvements are significant, YouTube channel Moore’s Law is Dead was previously much more optimistic about the performance gains, having previously reported that the new Ampere graphics could deliver up to a 3x to 4x uplift, according to Twitter user @jonastaghizadeh.

In our testing of the RTX 3080, we found that the RTX 3080 delivered solid performance uplift compared to the RTX 2080, but the gains weren’t quite as high as what Nvidia claimed. In Assassin’s Creed Odyssey, we found that the RTX 3080 was able to hit 61 FPS on the game on native 4K resolution when played at 60 FPS on Ultra High settings. This translates to a 30% performance boost compared to the RTX 2080 Super. Similarly, in the graphics-intensive Battlefield V, the game averaged 97 FPS in 4K at Ultra settings, a result that was 33% better than the RTX 2080 Super.

Because of the RTX 3080’s improved hardware, it performed its best — relative to the RTX 2000 series — when the resolution is pushed. At lower resolutions or game settings, we found little differences in performance, so 1080p gamers could skip this card and saving some money. This is a card that was designed to be pushed, as this is a GPU for 4K gaming at 60 FPS, according to Nvidia’s claims.

Moore’s Law is Dead reported that the RTX 3080 should put Nvidia in the performance lead once again against newer challengers from rival AMD. The RTX 3080 boasts nearly 30 teraflops of performance, compared with 12 teraflops on the Xbox Series X, which will ship with AMD’s RDNA 2 graphics architecture  When it comes to gaming, Nvidia’s RTX 3000 series cards will show that PC gaming still commands a performance lead over consoles.

Ahead of the card’s unveiling a leaked 3DMark Time Spy benchmark posted by HardwareLeaks from the RTX 3080 Ti (now known as the RTX 3090) showed the card had a 30% performance improvement over the RTX 2080 Ti. In a recent tweet, @kkatcorgi suggested a 20% generational performance uplift, though it wasn’t clear which benchmarking method or utility was used to obtain this figure. This could in turn lead to the 3090 gaining 40% better graphics performance than the 2080 Ti, according to Notebookcheck. Nvidia claimed that the RTX 3090 will be able to handle 8K gaming in 60 FPS.

For comparison, earlier leaked benchmarks for AMD’s Big Navi GPU, which uses the company’s RDNA 2 architecture, suggested that the card performs just 10% to 15% better than Nvidia’s RTX 2080 Ti. If accurate, this would mean that the RTX 3080 Ti could outperform AMD’s next-gen graphics for serious gamers, and place Big Navi’s performance more in line with the non-Ti variant of the RTX 3080.

While the generational improvement in performance is expected for a new GPU family, the new Ampere cards are also expected to perform significantly better at ray tracing, the hallmark feature of the RTX family and an area of struggle for the current RTX 2000 series. Ampere’s improvements with artificial intelligence and DLSS should help improve ray tracing performance, especially when games are rendered at higher resolutions.

Image source: Tweak Town

To achieve this level of performance, Nvidia was rumored to have included a new traversal coprocessor as part of the design of the RTX 3000 series, and the silicon’s function is to deliver better ray-tracing performance. This coprocessor would be the GPU’s ray tracing chip, according to Tweak Town, and could serve to unite the RT cores under one silicon.

It’s still unclear how this architecture would play out, and if Nvidia will offload all or some of the ray-tracing demands to this traversal coprocessor. Nvidia only claimed that its second-generation ray tracing GPU comes with enhanced RT and tensor cores. The RTX 3080 comes with 68 tensor cores, compared to just 46 on the RTX 2080. The RT cores are capable of 58 RT TFLOPS on the new RTX 3080, compared to just 25 RT TFLOPS on RTX 2080.

In our test of the RTX 3080 on Battlefield V, we found ray tracing capabilities to be far superior when compared to the RTX 2080 Ti. In the mission Tirailleur, we found that the scenes rendered beautifully, with nice reflections and shadows, and the game played at 55 FPS in 4K at Ultra settings. When DLSS was enabled, performance jumped to 69 FPS, compared to just 45 FPS on the RTX 2080 Ti.

Still, there were some performance bottlenecks when ray tracing was turned on. In Fortnite, for example, our review showed that the RTX 3080 was able to squeeze out an average of 23 FPS when played at 4K with the highest settings enabled. In 1440p (2K) resolution, the game averaged 53 FPS. Depending on the game, your monitor’s resolution, and what you hope to achieve, you may want to still leave ray tracing disabled at higher resolutions even on the RTX 3080.

Editors' Recommendations