Consider including 4070 Ti (non-Super) in initial testing results for 5070 Ti.
Consider including 4070 Ti (non-Super) in initial testing results for 5070 Ti.
Yeesh, the review sounds challenging. Thank you for putting it together. I’m sure it took a lot of time, but it was a solid read.
Good review.
I don't like the mixing of DLSS 2/3 results with straight raster on a single chart. It might confuse people into thinking the 5070Ti is significantly better than the 4090. Of course, you can explain it as much as you want, but many still overlook the unnecessary details.
Upscaling/DLSS/FSR/FG on one (or more) charts – Straight raster on another – This is the correct approach.
There is no testing on DLSS/FSR/FG in any of the charts here. If this isn't clear, let me know which ones are causing confusion. When discussing DXR with rasterization, the only charts combining these elements appear in the second set of ray tracing charts. Essentially, they're not prominently displayed. By then, it should be obvious what each chart represents.
And it's important to consider the broader context. While rasterization is still crucial and many RT games perform poorly, Nvidia delivers better upscaling quality, frame generation, and MFG. These aspects are combined into an overall performance metric.
Finding the ideal balance between these factors is a significant discussion point, which is why I've created separate charts for rasterization and RT. However, I also provide a comprehensive overview if you prefer. Choose what suits your needs.
When discussing the sensation of games with MFG or upscaling, you're actually experiencing them through play or based on observations from automatic benchmarks. You also ask about using game sequences unique to Tom, if applicable.
I swear I noticed some DLSS results in the bar charts. Not anymore, though. From a big-picture view, I'm okay as long as it's accurate and not misleading.
The ideal setup is having sufficient separate bar graphs so the reader can assess things themselves.
Go ahead and share your DLSS/FSR/FG graphs, but also include a baseline without any upscaling—just a standard raster graph that never shows upscaling examples.
From the 22 games examined in the review, only a few rely on a "canned" benchmark where I don't actually play (Black Myth Wukong, F1 24, and both Flight Simulator titles). The remaining ones are run manually, which means I can experience the game firsthand. It's challenging, but it offers a more genuine perspective. For my own use, I employed manual benchmarks and also tested the games outside of official testing. My goal was to grasp the overall feel and appearance rather than just focusing on performance metrics. The results are often unclear, varying from good to mediocre or even disappointing.