Nvidia has finally announced the RTX 4090 and RTX 4080 during the special GeForce Beyond broadcast. The cards are coming soon and are packed with new features. So much so that during the presentation Nvidia focused much less on raw performance and much more on the software stack that supports Ada Lovelace GPUs.
There are many unanswered performance questions about the RTX 4090 and RTX 4080, especially considering they are more expensive than the previous generation. Still, the cards are a sign of what’s to come in the future of PC graphics. Power definitely matters, but game reliability increasingly comes down to the software that new GPUs support. Nvidia is making that call, and it’s the right thing to do.
Where brute force ends
It’s easy to assume that you need to throw more force at the performance issue to increase frame rates, but the speed at which Nvidia moves doesn’t allow the hardware to catch up. If Jensen’s comments are anything to assume — where games become one massive simulation — you’d need a GPU across generations.
Currently, the highest quality rendering takes place offline. That is, a GPU works on rendering every frame, no matter how long it takes. Games do the same rendering, but it happens at runtime. The application (your game) is executing instructions while the rendering is taking place, and the rendering must be fast enough to have a playable frame rate.
The GPUs available to gamers are in many cases the same ones available to visual effects and animation studios, so this leap in fidelity cannot be solved by releasing new GPUs. There has to be a way to optimize runtime performance to have the same quality as an offline render. At least that’s the idea.
Nvidia achieves this through two main tools: Deep Learning Super Sampling (DLSS) 3 and Shader Execution Reordering (SER). Most Nvidia users are familiar with DLSS. It’s a supersampling tool that uses a lower internal resolution before scaling it up with AI to achieve your monitor’s native resolution. Fewer pixels to render means faster performance. Simple enough.
DLSS 3 aims to take that one step further. In addition to traditional AI upscaling, DLSS 3 generates completely unique frames that are not rendered by your GPU. It looks at past and future frames and can predict what your GPU would display in between. It should go without saying, but unique frames that your GPU doesn’t need to render are a tempting offer to boost performance.
SER specifically applies to ray tracing, which is necessary for a game that wants to be one big simulation. Nvidia says it reschedules shading workloads in real time, which can improve your frame rate by up to 25%. Nvidia equates SER with out-of-order execution for CPUs, which is quite a claim.
Typically, ray tracing does not work well with parallel processing. That is, each ray needs its own attention due to the semi-random nature of how light reflects. GPUs are really made for parallel processing, which is why ray tracing is so demanding these days. By optimizing the ray tracing instructions based on the available power, Nvidia hopes to increase the frame rates even further.
In the world of CPUs, there has been a lot of talk in recent years about the end of Moore’s law. GPUs and CPUs get to a point where it’s not as simple as packing more transistors into the same space, which was the idea behind Moore’s Law in the first place. And we are already seeing how software can fill in the gaps.
Of course, DLSS isn’t the same as native rendering – there’s a slight compromise in image quality – but it offers a much bigger performance boost than even a typical GPU generation. That is a difficult fact to deal with. A software feature, free to Nvidia RTX owners, provides more performance than a brand new GPU. In fairness, the same goes for AMD’s FidelityFX Super Resolution (FSR), especially in the most recent version.
We’re reaching a point where GPUs are defined by the software they support. SER and DLSS 3 are just the beginning of that. Obviously, Nvidia sees that the performance gains available through software optimizations can even outweigh generational improvements if there’s enough creativity, and that should mean higher frame rates for gamers even if they can’t afford a top-spec GPU.
Will it work?
While Nvidia’s claims about DLSS and SER make sense, that doesn’t mean the software combination is a winning lottery ticket. This is the same company that debuted ray tracing in 2018 to an absolutely horrendous lineup of games, and quite frankly, even ray tracing these days isn’t worth the performance shortfall in many cases. That could happen again with DLSS 3 and SER.
But I’m confident. There are definitely issues with the pricing of the RTX 4090 and RTX 4080, but the technology underneath is promising. DLSS now works wonders and hopefully Nvidia learned from the first version how the picture quality can weaken the experience. SER is also exciting, promising free ray tracing performance with no special rendering techniques.
However, my confidence is not concrete. It’s important to wait for the RTX 4090 and RTX 4080 to arrive before jumping to any firm conclusions. Either way, the future of PC graphics has been and will be determined by software, and Nvidia’s latest GPUs are proof of that.