After several months of development, the DLSS3 patch for Cyberpunk 2077 is now available to the general public. The press, including PCGH, have had access to a beta version since mid-October. So the final release was a long overdue step and will probably bring it Improvements in optics with – even if the crown jewel for the application, the ray tracing overdrive mode, is still missing.
RTX 3090 Ti versus RTX 4000
Nvidia does not miss the chance to drum for DLSS3. “Multiplying the frame rate” has been heard a number of times since the RTX 4090 was released. And the performance gains are ultimately impressive too. In addition, there is the reflex package in DLSS3 to minimize latencies. In Nvidia’s own benchmarks, this means that the RTX 4090 at least with DLSS performance in Ultra HD with Raytracing Ultra jumps from 41.8 without DLSS to 138 fps, i.e. 3.3 times as fast. With the RTX 4080 it is even 3.8 times as much, specifically from a rather unplayable 26.3 to 102.5 fps. The setting is four times as fast with an RTX 4070 Ti, which goes from a jerky 20.8 without DLSS to 84.4 fps with DLSS3. The RTX 3090 Ti, the former flagship for 2,249 (!) Euros, which was only released at the end of March 2022, looks comparatively old. 21.0 fps without DLSS and 57.5 fps with DLSS, version 2, mind you, are in Nvidia’s files. See the gallery below for quality mode benchmarks.
Compared to the RTX 4070 Ti, the RTX 3090 Ti looks pretty bad in fps. But not only that, Nvidia also draws comparisons in terms of energy efficiency, which should not flatter the once expensive RTX 3090 Ti and its owners. This is the power consumption of the RTX 3090 Ti in Cyberpunk 2077in exactly the above scenario (Ultra HD, Raytracing Ultra, DLSS performance with DLSS2 and 3) comparatively high: Nvidia measures 388.1 watts on a Core i9-12900K with 32 GiB RAM, the RTX 4090 is 310.2 watts in comparison , the RTX 4080 at 256.6 watts and the RTX 4070 Ti at 240.4 watts. Nvidia has already focused on the energy efficiency of Ada GPUs.
At that time it was explained that the GPU Boost algorithms on RTX 30 GPUs increased the clock frequency until the performance cap, in this case the TGP, was reached. With Ada GPUs, you first come up against other limits, for example the maximum clock frequency or the maximum voltage. One is able to “reduce the overall performance in order to save energy”. As an example, the scenario is given that the GPU is limited by the CPU or the GPU load is comparatively low. Then the GPU boost clock frequencies still reach the maximum frequency of the graphics processor, but the power consumption is sometimes further below the TGP, according to Nvidia.
Overall, Nvidia makes the RTX 3090 Ti, the former flagship, look comparatively old. How do you like the comparisons? Use the comment function and let us know what you think. You must be logged in to PCGH.de or the Extreme forum to comment. If you don’t have an account yet, you can consider registeringwhich brings many advantages.