Nvidia's next-gen RTX 30 cards are first to use new Ampere architecture
New graphics cards offer up to twice the performance of Turing-based models
Nvidia has announced its next-generation RTX 30 series graphics cards, based on the Ampere architecture. The RTX 3070, RTX 3080 and RTX 3090 feature improvements in rasterised and ray traced 3D rendering, as well as new machine learning optimisations.
"Today's launch of Nvidia Ampere GPUs is a giant step into the future," said CEO Jen-Hsun Huang.
"The work of thousands of engineering years, the GeForce RTX 30 Series delivers our greatest generational leap ever. NVIDIA RTX fuses programmable shading, ray tracing and AI for developers to create entirely new worlds. Twenty years from now, we'll look back and realise that the future of gaming started here."
According to Huang, the new cards usher in several improvements over soon-to-be previous generation. In terms of software, Nvidia has been investing in machine learning algorithms to lower the load on the GPU front-end. The deep learning super sampling (DLSS) 2.0 process upscales graphics using low-resolution inputs; the new cards will use this to render landscapes. Unlike DLSS 1.0, which Nvidia launched two years ago, the new version is trained on non-game-specific content, which can work across multiple titles.
The most powerful of the new cards is the RTX 3090, which boasts 50 per cent improved performance over the Titan RTX. The three-slot RTX 3090 sports 24 GB of RAM, and Nvidia claims it can achieve 8K resolution at 60fps 'across many top games'. Other specifications include 19.5Gbps memory bandwidth, 1,695MHz boost clock and 350W max power consumption.
The RTX 3090 is priced at whopping $1,499; higher than the launch price of the $1,299 RTX 2080 Ti. It will go on sale starting 24th September.
The next in the list, RTX 3080, is said to be twice as fast as the RTX 2080. Among its most notable features are 10GB of GDDR6X memory, with 760Gbps bandwidth, 1,710MHz boost clock, 19Gbps memory bandwidth, and 320W max power consumption. It will ship with a dual 8-pin adapter for customers whose PSUs do not feature a smaller 12-pin cable.
The RTX 3080 will go on sale on the 17th September, at $699.
Finally is the RTX 3070, which sports 8GB of GDDR6 memory, a 256-bit memory bus, 1,725MHz boost clock, 14Gbps memory bandwidth and 220W max power consumption. Nvidia claims this card is 60 per cent faster than the RTX 2070 and is suitable for 4K and 2560 x 1440 gaming.
The RTX 3070 is priced at $499, the same as the launch price for the RTX 2070 Super. The card is expected to hit shelves in October.
Nvidia says the Ampere architecture brings several key new features to the GeForce RTX 30 series. First introduced this spring as part of the A100 accelerator, Ampere features 28 billion transistors on a card, compared to Turing's 19.8 billion transistors. It also supports high-bandwidth memory, second-gen ray tracing cores (plus concurrent ray tracing, shading and compute on the RTX 30 Series) and multi-instance GPU partitioning.
More on Chips and Components
TSMC expects 3nm chip mass production in the second half of 2022
3nm technology will be 30 per cent more energy efficient and 15 per cent faster than the 5nm tech, TSMC says
US sanctions force Huawei to stop making Kirin chips
Huawei can no longer produce the chips it needs for its high-end smartphones with features like 5G and AI, due to US government sanctions
Intel's 7nm chips will not arrive until late 2022
Rivals AMD and Nvidia have already started taking advantage of 7nm technology to produce more efficient chips
TSMC to begin 3nm mass production in 2021, report
The chip maker is also rumoured to have made major breakthrough in 2nm chips
Apple ditches Intel to switch Mac computers to its own ARM-based processors
The first Macs with Apple chips will arrive by the end of the year
Most read
