Alexander "peter bjorn" Fadin Disappointing news is starting to surface ahead of the February 2025 release of the RTX 5070 GPU. The new chip may prove to be less arbitrary than its predecessor, the RTX 4070 Super, according to official NVIDIA specifications, particularly without Multi-Frame Generation (MFG) technology.
Compared to the RTX 4070 Super, the RTX 5070 has 14% fewer CUDA cores (6144 vs. 7168) and 14% fewer Texture Mapping Units (192 vs. 224). This reduction in the number of cores and units can result in lower performance under standard workloads if MFG, which helps boost frame rates with AI, is not utilized.
Despite the fact that the RTX 5070 will be equipped with 12GB of GDDR7 memory and it consumes 250W (which is 14% more than the RTX 4070), the potential performance in its raw form, that is, without enhancement technologies, may be inferior to its predecessor. In this context, many experts express doubts that the new chip will meet players' expectations.
Image: tech4gamers.com
If we take as an example more productive cards, such as RTX 5080, it will demonstrate only 8-9% performance gain compared to RTX 4080 Super. That is, even the older models don't show significant performance gains without considering MFG or DLSS technologies.
For many gamers who prefer consistent raw performance, the lack of significant performance gains in the RTX 5070 will be a disappointment. While MFG will help improve game smoothness, not all gamers are in the mood to rely on the technology.
The RTX 5070 will go on sale on February 20, 2025 with a price tag of $549, but it's already becoming clear as new specs and benchmarks are released that the RTX 50-series may not bring the expected jump in performance.
Main image: reddit.com
0 comments