NVIDIA’s DLSS Multi Frame Generation technology is officially limited to GeForce RTX 50 series graphics cards, but modders have now managed to run versions of the feature on three older GPU generations.
Recent community projects have brought native DLSS frame generation code to RTX 40, RTX 30 and even RTX 20 series cards. These projects differ from older workarounds that relied on AMD’s frame generation technology. Instead, they attempt to run NVIDIA’s own DLSS Multi Frame Generation pipeline on hardware that is not officially supported.
The results are promising, but compatibility remains inconsistent and the mods can break after game, driver or DLSS updates.
| GPU generation | Architecture | Current status | Tested capability |
|---|---|---|---|
| RTX 40 series | Ada Lovelace | Most advanced | Multiple MFG modes available |
| RTX 30 series | Ampere | Experimental | 2X, 3X and 4X in supported cases |
| RTX 20 series | Turing | Early testing | Frame generation enabled in limited testing |
RTX 40 series support is currently the most developed
Ada Lovelace cards are the furthest along because their hardware is closer to the RTX 50 generation.
One project provides a broad DirectX 12 and Vulkan solution for games that already support DLSS Frame Generation through NVIDIA’s Streamline framework. It can enable fixed Multi Frame Generation multipliers and also includes experimental dynamic modes.
Other implementations take different approaches, including rebuilding Blackwell targeted DLSS kernels so they can operate on Ada GPUs.
The latest versions already include compatibility fixes, but the experience is still not equivalent to official support. Games may behave differently depending on their DLSS implementation and software stack.
RTX 30 cards can now run native DLSS frame generation code
Ampere support is a more significant technical step because RTX 30 GPUs were never designed with official DLSS Frame Generation support.
A newer project adds an SM86 backend for NVIDIA’s DLSS Multi Frame Generation runtime. SM86 refers to the CUDA compute capability used by Ampere GPUs.
The implementation has been tested on an RTX 3080 Ti and can expose 2X, 3X and 4X frame generation modes in supported situations.
However, compatibility is far from complete. Reports already include missing frame generation settings, crashes and visual artifacts in some games.
That means RTX 30 owners should view the project as experimental rather than a reliable way to gain RTX 50 class features.
Even RTX 20 series GPUs have entered testing
Turing cards are now part of the same experiment.
A separate project adapts the Ampere implementation for SM75, the compute capability associated with RTX 20 series hardware. This requires rebuilt kernels and modified architecture checks.
Testing has so far been much more limited. An RTX 2060 Max Q was able to enable DLSS Frame Generation in The Witcher 3 using DirectX 12, but that single example does not establish broad compatibility.
Performance, image quality and stability across other RTX 20 models remain uncertain.
All of these projects depend on software components that NVIDIA still controls, including NGX, Streamline and closed source DLSS runtime elements. Previous updates have already disrupted earlier unlock methods before modders found new workarounds.
Future driver or DLSS changes could therefore disable some of these techniques again.
For now, the work shows that the software restrictions around Multi Frame Generation can be bypassed to some extent. It does not mean older RTX cards will deliver the same experience, efficiency or reliability as RTX 50 hardware running the feature officially.
Discussion (0)
Be the first to comment.