Nvidia has once again stirred excitement and skepticism in the gaming and tech world.
At CES 2025, CEO Jensen Huang made a statement that set the internet ablaze: the upcoming RTX 5070 GPU will deliver the performance of the flagship RTX 4090 for just $549. The promise hinges on Nvidia’s groundbreaking DLSS 4 technology, which leverages AI-powered Multi Frame Generation. But can this claim really hold water? Here’s a deep dive into what this could mean for gamers and the future of GPU performance.
Table of Contents
The Technology Behind the Claim: DLSS 4 and Multi Frame Generation
At the heart of Nvidia’s bold assertion lies its next-generation Deep Learning Super Sampling (DLSS) 4. This AI-driven technology represents a significant leap forward in rendering performance, but it comes with its share of controversy.
How DLSS 4 Works
DLSS 4 introduces Multi Frame Generation, allowing the GPU to generate up to three additional frames for every traditionally rendered frame. In essence, this means a single frame rendered by the GPU is amplified into four visible frames, dramatically increasing frame rates.
Some key advancements in DLSS 4 include:
- Transformer Model Technology: Nvidia claims this new model is 40% faster and uses 30% less VRAM compared to DLSS 3.
- “Predicting the Future”: Unlike earlier iterations, which interpolated frames based on past data, DLSS 4 predicts upcoming frames, reducing latency.
- Massive Frame Rate Boosts: Nvidia demonstrated in games like Cyberpunk 2077, where DLSS 4 pushed frame rates from 27fps (without DLSS) to 243fps, an almost 8x increase.
While these enhancements are undeniably impressive, they’ve reignited a long-standing debate among gamers over so-called “fake frames” — frames generated by AI rather than traditional rasterization.
The “Fake Frames” Debate: Perception vs. Performance
Nvidia’s reliance on AI-generated frames has been a contentious issue since the introduction of DLSS 3. Critics argue that these frames are not “real” because they aren’t rendered in the conventional sense. Instead, they are created algorithmically to smooth gameplay and improve visual fluidity.
The Pros and Cons of AI-Generated Frames
- Advantages:
- Higher Frame Rates: DLSS 4 offers unprecedented frame rate improvements, even on mid-tier GPUs like the RTX 5070.
- Enhanced Visual Quality: AI-generated frames can make demanding games like Alan Wake 2 and Cyberpunk 2077 playable with full ray tracing.
- Efficient Resource Use: By offloading work to Tensor Cores, DLSS frees up the GPU’s primary rendering capabilities.
- Drawbacks:
- Latency Concerns: While Nvidia claims latency increases are minimal (around 6ms with Multi Frame Generation), gamers sensitive to input lag may still notice the difference.
- Subjective Experience: Even with higher frame rates, some players report that games feel slower, especially when transitioning from native frame rates like 240fps to AI-generated rates.
Why It Matters
For casual gamers, the distinction between “real” and “fake” frames may be negligible. However, competitive gamers and purists often prioritize input responsiveness and native rendering quality, where the RTX 5070 will likely fall short of the RTX 4090.
Check also: AMD’s Strix Halo Leak: The Future of Laptop CPUs Unveiled
Breaking Down Nvidia’s Performance Claim
Nvidia has clarified that the RTX 5070 won’t deliver RTX 4090 performance in all scenarios. Instead, the comparison applies primarily to DLSS-enabled games. According to Nvidia’s Lars Weinand, the claim refers to frame rate parity, not raw rasterization performance.
DLSS Dependency
The RTX 5070’s ability to achieve RTX 4090-like performance relies heavily on DLSS 4 adoption. Without DLSS support, the GPU’s raw rasterization power will likely fall far short of the RTX 4090.
Real-World Scenarios
- DLSS-Enabled Games: In titles like Cyberpunk 2077, the RTX 5070 can deliver comparable frame rates to the RTX 4090 by leveraging DLSS 4’s Multi Frame Generation.
- Non-DLSS Games: Pure rasterization tests will reveal the RTX 5070’s true limitations, making it more comparable to its direct predecessors (e.g., the RTX 4070).
