Modders have coaxed Nvidia’s Multi Frame Generation to work on RTX 40-series cards, making one Blackwell selling point look less exclusive.
According to Tom’s Hardware, the unofficial mod from developer dashdogy enables Multi Frame Generation on Ada Lovelace hardware, despite Nvidia marketing the feature as a major attraction of its RTX 50-series GPUs. Tom’s Hardware tested the mod in Cyberpunk 2077 and found that Multi Frame Generation genuinely works on RTX 40-series cards.
Increasing the MFG multiplier beyond the officially supported 2X level on Ada noticeably improved perceived smoothness during testing. The mod produced the same kinds of visual artefacts seen on RTX 50-series hardware as additional generated frames were added.
At MFG 4X and above, both generations displayed visual junk near the bottom of the image, suggesting Ada really was running the same sort of frame-generation workload rather than enthusiastically pretending. Tom’s Hardware tested cards ranging from the RTX 4090 down to the 12GB RTX 4070.
The test used Cyberpunk 2077 with maximum raster settings, path tracing, DLSS Super Resolution, Ray Reconstruction and MFG 4X enabled. Tests were carried out at 1080p with DLSS Balanced, 2560×1440 with DLSS Performance and 4K using DLSS Ultra Performance.
At 1080p, every RTX 40-series card tested stayed below Tom’s Hardware’s roughly 60ms average input-latency threshold with MFG 4X running by itself. Blackwell still had an advantage, with even the RTX 5070 delivering lower latency than the RTX 4090 in some configurations.
At 2560×1440, every Ada card except the RTX 4070 remained within the publication’s acceptable latency range when running modded MFG 4X alone. Adding an experimental DLSS 5 mod made matters considerably messier, pushing latency beyond 60ms on every Ada card except the RTX 4090.
The RTX 4080 Super and RTX 4080 remained only slightly beyond the threshold, while weaker cards required compromises such as reducing ray tracing or abandoning path tracing. The results became stranger at 4K.
Ada cards unexpectedly showed lower input latency at 4K with DLSS Ultra Performance than at 1440p with DLSS Performance, even though both modes used the same internal rendering resolution.
Tom’s Hardware repeated the measurements and found the behaviour reproducible, although it could not explain the latency bump at 1440p. The RTX 4090 even moved ahead of the RTX 5080 at 4K in some MFG 4X tests, while the RTX 4070 Ti Super appeared to benefit from its 16GB of VRAM.
Plenty of catches remain before RTX 40 owners start congratulating themselves for skipping an upgrade. The implementation remains an unofficial mod in active development, and one newer version intended to remove the ReShade requirement crashed Cyberpunk 2077 during testing.
Blackwell retains a measurable latency advantage, particularly when more demanding AI rendering techniques are piled on top of Multi Frame Generation. Still, the results make Nvidia’s previous presentation of MFG as a Blackwell-specific attraction look slightly awkward.
Tom’s said its testing suggested there was no obvious technical reason preventing Nvidia from bringing MFG to Ada officially.
Nvidia is already preparing DLSS 5 support for RTX 40-series cards later this year, although it has not announced official Multi Frame Generation support for Ada. With graphics card prices remaining unpleasant and gamers increasingly inclined to hang on to older hardware, an official switch could extend the useful life of RTX 40 cards without requiring another visit to Nvidia’s increasingly ambitious price list.
