Graphics card manufacturer NVIDIA has been stealing the tech headlines this past week with the unveiling of its new series of Turing-based graphics cards which include what is dubbed the holy grail of graphics – real-time ray tracing.
First teased at the Game Developers Conference back in March with a “Star Wars” clip, ray tracing is seen as the pinnacle when it comes to lighting computer-generated environments. The computer maps millions of light rays through an image to create natural and realistic shading, reflections, and depth-of-field.
Used extensively in major blockbuster movies which have the time and money to do so, the tech has been far too demanding to be used in real time for things like video games because graphics cards just haven’t been powerful enough – that’s why there’s a difference between pre-rendered cut scenes and in-engine gameplay. An excellent explanation of ray tracing and its uses was done back in April by Engadget and you can see the video here.
Things are now taking a big step forward as NVIDIA used the SIGGRAPH conference last week to tout its professional-level Quadro RTX cards offering “capabilities that weren’t expected to be available for at least five more years” according to company exec Bob Pette. The cards are reportedly at least six times more powerful than current Pascal-based cards and boast not only dedicated real-time ray tracing cores but tensor core-processed AI routines and advanced shading.
This allows designers and artists to interact in real time with complex designs and visual effects in ray-traced photo-realistic detail and significantly cuts down on studio rendering workloads – leading to significant time and cost savings for artists, animators and digital production workers.
It also allows for the rendering of completely computer-animated clips like these below not from a server farm taking days to churn out the result, but from a single server with the commonly used Unreal Engine 4 and just one or two of these $4-10K graphics cards allowing animators to change elements in these scenes on the fly as they’re processed in real time:
However what gamers want to know about is the consumer level cards. NVIDIA is set to make the official announcement regarding those, its first Turing-based GeForce video cards, on Monday ahead of GamesCom. The successor to the current most powerful cards on the market including the GTX 1080Ti and the Titan V, the RTX 2080 line is changing things up – and surprisingly it looks like they may be launching both the regular 2080 and the 2080Ti close together if not simultaneously.
Specs have begun to leak out over the weekend with VideoCardz reporting that the 2080Ti boasts 4352 CUDA Cores and 11Gb of GDDR6 Memory, while power consumption is only a little higher at 285W (compared to 250W for the 1080Ti). The cards are capable of 8K resolution at 60FPS and have 616 Gb/s bandwidth – a big boost from the 448 Gb/s of the previous generation.
Specifics regarding the cards will be unveiled during Monday’s presentation. Microsoft has reportedly created a new DirectX Raytracing (DXR) API to compliment Nvidia’s RTX work and real-time ray tracing is expected to be available to Unreal Engine developers later this year.
Cost is also unknown at this point, but with the cryptocurrency mining boom over (Bitcoin is now trading at just 30% of its peak price back in late 2017) there isn’t expected to be a shortage of cards like there has been in the past year or so due to cryptocurrency miners buying up said cards.