Nvidia introduces GeForce RTX 3070, 3080 and 3090 GPUs
Wednesday, September 2, 2020 | Written by Jim Thacker
Nvidia has introduced the GeForce RTX 3070, GeForce RTX 3080 and GeForce RTX 3090, the next-generation gaming GPUs of the GeForce RTX 30 series based on the new Ampere GPU architecture.
The company positions the Ampere cards as a significant step up from their predecessors, even describing the mid-priced RTX 3070 as being faster than its previous flagship, the RTX 2080 Ti.
We don't normally cover consumer GPUs on Greenscreen, but as the lines between gaming and workstation cards become increasingly blurred and Nvidia now includes Arnold, Blender and V-Ray in its GeForce performance graphs, we've rounded up the key specifications for CG- Artist below.
|Nvidia GeForce RTX GPU specifications|
|RTX 3090||RTX 3080||RTX 3070||RTX 2080 Ti|
|Tensor cores *||328||272||184||544|
|RT cores* *||82||68||48||68|
|Base clock (GHz)||1.40||1.44||1.50||1.35|
|Boost clock (GHz)||1.70||1.71||1.73||1.64|
FP32 (Tflops) *
|GPU memory||24 GB|
|Release date||September 2020||September 2020||October 2020||September 2018|
|$ 1,499||$ 699||$ 499||$ 1,199|
* Information comes from third party websites
Significant but iterative improvements in GPU processing power
From a GPU computing perspective, Ampere doesn't look too different from Turing, his predecessor: it's an iterative improvement, not a revolution.
Unlike Turing and its dedicated RT ray tracing cores, Nvidia didn't introduce any new types of hardware this time around, although the designs of the existing cores have been updated.
Nvidia describes the RT cores of the second generation of Ampere cards with 1.7 times the throughput of their predecessors. and their third generation machine learning tensor cores as a 2.7X improvement.
When it comes to DCC apps, Nvidia shows that second generation RT cores roughly double ray tracing performance with key renderers like Arnold from Autodesk, V-Ray from Chaos Group, and Cycles from Blender.
However, there are no details on how this number was found or how it is split between individual cards. So we're inclined to wait for some more conventional benchmarks.
The number of standard CUDA cores has also increased, as all Ampere cards have a higher theoretical FP32 processing power than the previous top-of-the-range GeForce RTX 2080 Ti.
Nvidia describes all Ampere cards – even the GeForce RTX 3070 for 499 US dollars – as faster than the RTX 2080 Ti and the RTX 3090 as "up to 50% faster (at 4K)" than the Titan RTX, their hybrid workstation / Gaming card.
Increased graphics memory capacity and NVLink support for GPU rendering
A new technology that was introduced in the Ampere cards is the GDDR6X memory, which is intended to increase the memory bandwidth compared to the GDDR6 memory of the previous generation of cards.
The graphics memory capacity has also been increased. The GeForce RTX 3090 has 24 GB of integrated RAM.
This corresponds to the current Quadro RTX 6000 workstation card from Nvidia, which Jason Lewis called the "sweet spot" for serious GPU rendering work in his 2020 group test for professional GPUs.
The RTX 3090 also supports NVLink, which makes it possible to pool GPU memory between cards: a feature that seems to be geared more towards CG artists than gamers.
However, power consumption increases, which is reflected in the RTX 3090's TDP of 360W – almost 100W higher than the 2080Ti's – so electricity may incur production costs.
Other major display standards and APIs
In addition, the new cards support PCIe 4.0 – they should work with "all PCIe 3.0 and 4.0 CPUs (and) motherboards" as well as the HDMI 2.1 and DisplayPort 1.4a display specifications.
In terms of graphics APIs, all new cards support DirectX 12, OpenGL 4.6 and Vulkan RT.
Prices and release dates
The GeForce RTX 3070 ships in October 2020 with an MSRP of $ 499. The RTX 3080 ships on September 17, 2020 with an MSRP of $ 699. The GeForce RTX 3090 ships on October 24, 2020 with an MSRP of $ 1,499.
For more information on the GeForce RTX 30 series cards, visit the Nvidia blog
Tags: 8nm, AI, Ampere, Arnold, Autodesk, Benchmark, Blender, Boost clock, CG, Chaos Group, clock rate, comparison table, computing power, CUDA core, cycles, DCC, FP32, game card, GDDR6, GDDR6X, GeForce RTX, GeForce RTX 2080 Ti, GeForce RTX 30 series, GeForce RTX 3070, GeForce RTX 3080, GeForce RTX 3090, GPU, GPU computing, GPU playback, graphics card, graphics memory, machine learning, NVIDIA, NVLink, performance, price, processor speed , Ray tracing, release date, RT core, specs, specs, TDP. Power consumption, tensor core, Titan RTX, V-Ray, VRAM, workstation card