Yesterday was a pretty big day for product announcements but by far the most exciting stream we watched was from Nvidia.

Late of Tuesday afternoon, Nvidia chief executive officer Jensen Huang hosted a press conference, in his kitchen.

Rather than whipping out a tiny whisk (or one of the thousands of silicon spatulas he seemingly has), Huang showed off Nvidia’s RTX 30 series of graphics cards. While the performance gains seem good on paper – twice the performance, 1.9 times the power efficiency compared to previous generations – visually Nvidia has taken a veritable leap from the RTX tech we can experience right now.

The RTX 30 series are driven by Nvidia’s Ampere architecture and according to the firm the next generation of dedicated ray tracing cores and improved CUDA performance speed up rendering time by twice as much as previous generations.

But what does this really mean? You can see an example of this improved performance in the video below.

The video above is a follow up to a render shown back in May titled Marbles. That video was rendered in real time at 720p 25fps with deep learning super sampling bumping that up to 1080p. This was however rendered on a Quadro RTX 8000, a card that isn’t designed for the mainstream.

The video above is what we’d call a generational leap. Marbles at Night is rendered in real time at 1440p 30fps on Nvidia’s Ampere architecture. It’s rather astounding to see this tech in action.

Another very interesting feature present in the RTX 30 series is RTX IO.

Together with Microsoft’s new DirectStorage for Windows API, RTX IO is able to offloads CPU work onto the GPU. The result is faster game loading times, improved frame rates and being able to pick up exactly where you left off in a game.

“Nvidia RTX IO plugs into Microsoft’s upcoming DirectStorage API, which is a next-generation storage architecture designed specifically for gaming PCs equipped with state-of-the-art NVMe SSDs, and the complex workloads that modern games require. Together, the streamlined and parallelized APIs, specifically tailored for games, allow dramatically reduced IO overhead and maximize performance/bandwidth from NVMe SSD to your RTX IO-enabled GPU. Specifically, NVIDIA RTX IO brings GPU-based lossless decompression, allowing reads through DirectStorage to remain compressed while being delivered to the GPU for decompression. This removes the load from the CPU, moving the data from storage to the GPU in its more efficient, compressed form, and improving I/O performance by a factor of two,” explains Nvidia.

If this sounds familiar it’s because both PlayStation and Xbox have touted a similar feature in the next generation of consoles, though this is specifically for PCs.

Three cards

During the stream Huang revealed three new GPUs:

  • RTX 3070
  • RTX 3080
  • RTX 3090

The RTX 3070 is the “sweet spot” and according to Nvidia it beats out its own RTX 2080 Ti in performance numbers while maintain a price point that starts at $499 (~R8 324 not accounting for import tax, fees etc.). Should that price point stay at around that mark when the cards hit South Africa (we have no faith that it will) that could mean you have better performance than a R26 000 GPU for a fraction of the price. That’s rather tempting.

The RTX 3080 is next in the line-up and to our mind the card that many gamers will be dreaming about.

The card is designed for 4K gaming and “consistently delivers more than 60 fps at 4K resolution – with RTX ON” according to Nvidia. That is most certainly a claim worth investigating and if it holds water, AMD best start working on a response, if it isn’t already.

The RTX 3080 is also said to be a faster and better than the RTX 2080Ti with prices starting at $699 (~R11 660 not accounting for import tax, fees etc.) for the RTX 3080.

Finally, the BFGPU or the Big Ferocious GPU – the RTX 3080.

To quote Huang, this thing is a beast, it packs in 24GB of GDDR6X running at 19.5Gbps, runs 30 degrees C cooler than the Titan RTX and, it runs games at 8K 60fps.

This alone has us curious because Nvidia doesn’t mention at what graphics preset this was accomplished at. While this claim is very impressive we’d like to see what 8K 60fps looks like from a  technical stand point.

The RTX 3090 will carry a starting price of $1 499 (~R25 006 not accounting for import tax, fees etc.).

Each of these cards will make use of a new 12-pin PCIe power connector, according to Gadgets 360, an adapter will ship with the cards so you don’t have to buy a new power supply.

That having been said, you may want to consider a power supply upgrade.

The RTX 3090 will reportedly draw 350W, the RTX 3080 will reportedly draw 320W while the RTX 3070 will draw a reported 220W.

So when can we expect these cards? Unfortunately we don’t have local release dates as of time of writing so we’re basing this off Nvidia’s website

The RTX 3070 will only be available in October, while the RTX 3080 will be available from 17th September with the RTX 3090 coming to market on 24th September.

As soon as we have local pricing and availability we’ll be sure to share that information.