Nvidia announces RTX 4090 and 4080 graphics cards, DLSS 3
September 20, 2022
Nvidia announced its latest generation of GeForce RTX graphics cards at its GTC AI conference. The Nvidia RTX 4090 and RTX 4080 are the first ‘Ada Lovelace’ series GPUs, and they’re expected to be up to twice as fast as their last-gen counterparts in pixel-based games — and up to four times faster in ray games. .
The RTX 4090 24GB will be available on October 12 and costs $1599, while the RTX 4080 16GB arrives in November and costs $1199. There’s also an RTX 4080 12GB model that costs $899. The 4090 and 4080 are said to be two to four times faster than their Ampere counterparts, which are the 3090 Ti and 3080 Ti respectively.
Note that Nvidia RTX 30-series cards will remain on sale for the time being, filling the bottom end of the market until the lower-tier RTX 40-series GPUs arrive.
So how are these GPUs so fast? Well, the 40-series GPUs use TSMC’s “4N” process and feature up to 76 billion transistors and up to 18,000 CUDA cores, which is 70% more than in the last generation of Ampere. The new process allows the build to be much more power efficient as well, although we expect the flagship cards to be as power hungry as rumored – you just get a ton of extra performance to make up for that in terms of efficiency.
The new streaming multiprocessor uses a new technique, Shader Execution Re-engineering (SER), which Nvidia says delivers “two to three times speedup” for ray tracing and a 25% improvement for ray tracing. pixelated games. SER works by dynamically rescheduling shader workloads to better utilize GPU resources. There are similar advancements in dedicated ray tracing silicon, with a doubling of ray triangle intersection throughput, a new opacity micromap engine that doubles the ‘alpha test geometry ray tracing’ speed and a micromesh engine that ‘increases geometric richness without the cost of building and stocking the BVH’. There is also a new, more powerful Tensor core for AI tasks.
It’s also the first card to support Nvidia’s DLSS 3, a new technique that generates entirely new frames – essentially adding interpolated frames between the ‘real’ ones to dramatically increase frame rates. This has a detrimental effect on input latency, so it is combined with Reflex to reduce input latency as much as possible. This reduces the load on the CPU and GPU, so you actually see the frame rate benefits in CPU and GPU limited games – although the input latency penalty means it will less useful for, say, competitive FPS games.
Nvidia demonstrated that Cyberpunk 2077 ran at ~22fps at 4K with RT on and DLSS off, then around 100fps with RT on and DLSS 3 on – a massive speedup even though it’s a selected demo. The company also released a video demo of Flight Sim 2020, with the game running at ~60fps with RT+DLSS disabled and ~135fps with RT+DLSS 3 enabled – good proof that the DLSS 3 technique alleviates the limitations of the processor.
In addition to upgrading existing games with RTX and DLSS 3, Nvidia also announced a “new” title: Portal RTX. This remaster, 15 years after the original release, looks impressive from the short trailer. It will be released for free to portal owners in November. The mod was created in what Nvidia calls “RTX Remaster”, an application that allows you to produce mods for a wide range of games that add ray tracing and AI-enhanced textures – very neat.
Elsewhere, Nvidia has announced Racer X, a “simulated world” with incredible realism – think their marble demo has gone 11.
It was an impressive presentation from Nvidia, and we can’t wait to test both the performance and next-gen features. It will be fascinating to see if DLSS 3 lives up to Team Green’s lofty promises…and if these new cards will actually be available at the advertised prices if performance is warranted. We’ll share more information when we have it!