When NVIDIA originally unveiled DLSS Frame Generation with the GeForce RTX 40 Series GPUs (codenamed “Ada Lovelace”), it noted that this technology couldn’t come to the GeForce RTX 30 Series “Ampere” GPUs because it required a specific hardware block built into the RTX 40 series, known as the “Optical Flow Accelerator.” However, the new Multi-Frame Generation debuting as part of DLSS 4 no longer uses that block and is instead performed in GPU compute.
Given that, it makes sense that DLSS Frame Generation should find its way to the GeForce RTX 30 series now, right? Digital Foundry‘s Alex Battaglia posed that exact question to NVIDIA’s Vice President of Applied Deep Learning Research, Bryan Catanzaro, and props to Alex for what is honestly a pretty hardball question. Catanzaro fielded it, though, saying:
I think this is primarily a question of optimization, and also engineering, and then the ultimate user experience. We’re launching this frame generation, the best multi-frame generation technology with the 50 series, and we’ll see what we’re able to squeeze out of older hardware in the future.
In other words, he’s not promising that frame generation will come to the GeForce RTX 30 series, but he’s also not saying that it won’t. In fact, if you read between the lines a little bit, what he’s actually saying is that the “Ampere” GeForce cards absolutely could get DLSS FG—it’s just down to NVIDIA’s higher-ups to decide that needs to happen.
Catanzaro is an engineer and works in development, not in PR, so his non-committal answer is completely understandable. Indeed, despite that some customers would undoubtedly be pleased to get DLSS FG support on their older GPUs, it could be a bad PR move for NVIDIA; it might make the early RTX 40 series restriction seem artificial and forced.
That’s purportedly not the case given the technology changes, but that nuance could be lost in internet arguments. It’s worth noting that both NVIDIA’s rival AMD and even third-party apps like Lossless Scaling have already demonstrated that effective frame generation—even multi-frame generation—can work without dedicated hardware. These solutions may not be as polished as NVIDIA’s, but they work on a wide range of GPUs.
Digital Foundry‘s interview with Catanzaro has many other interesting insights, like details on what spurred the move to a transformer architecture for DLSS as well as more information on the revised version of NVIDIA Reflex launching with these cards. If you’re interested, we encourage you to check it out using the embed above.