When Should You Use DLSS 4.0 Instead Of DLSS 4.5?

Summary

The discussion delves into the nuances of DLSS 4.5, with Nvidia's recommendation to use it for performance and ultra-performance modes, and the question of whether future updates will improve quality mode. Experts explain that the higher performance cost of DLSS 4.5 means it benefits performance-heavy modes, and that improvements to image quality are reaching diminishing returns. With GPUs like the RTX 5090, users are often better off sticking with DLSS 4. Overall, DLSS is advancing, but its impact on older GPUs is evident, pushing the focus to enhancing performance for modern systems, even at lower resolutions.

Takeaways

• 😀 DLSS 4.5 is recommended by Nvidia primarily for performance and ultra-performance modes, with users on quality or balance mode advised to stick with DLSS 4.0.
• 😀 DLSS 4.5's higher performance cost, due to better anti-aliasing and reconstruction techniques, makes it less suitable for quality or balance modes as it consumes a larger portion of frame time.
• 😀 Nvidia's emphasis on DLSS 4.5 for performance modes isn't about quality degradation but the additional cost to run the improved model.
• 😀 It's unlikely that a future DLSS update will improve the quality mode, as DLSS 4.5 is designed to prioritize higher performance, especially on modern GPUs.
• 😀 As tensor core technology improves in future GPUs, the performance cost of running advanced DLSS models may increase, leading to higher image quality, but older GPUs may struggle more.
• 😀 DLSS and similar upscaling technologies like FSR and XCSS will continue to become more demanding over time, but this is expected as the visual quality of games improves.
• 😀 5090 users may not need to stick to DLSS 4.0 in quality mode since DLSS 4.5 already looks great in quality mode for most objects, though using quality mode in general could be overkill for many games.
• 😀 With advancements in GPU and tensor core capabilities, the focus is now shifting toward using these tools for higher quality options without just increasing native resolution.
• 😀 By DLSS 4, the image quality has already reached a high level, making future improvements feel subtle, especially for more aggressive modes like performance and ultra-performance.
• 😀 DLSS continues to impress with its ability to extract high-quality visuals from very low internal resolutions, such as 480p, making it usable even for lower-end GPUs like the RTX 2070.
• 😀 Nvidia's work on improving image quality across DLSS presets, especially in ultra-performance mode, shows how even older hardware can benefit from advancements in AI-driven rendering technologies.

Q & A

Why is Nvidia recommending DLSS 4.5 only for performance and ultra-performance modes?

Nvidia recommends DLSS 4.5 for performance and ultra-performance modes because the higher computational cost of running DLSS 4.5 in quality or balance modes results in less performance improvement. The additional anti-aliasing and reconstruction techniques in DLSS 4.5 increase the base performance cost, making it less effective in modes that demand higher image quality.

Do you expect Nvidia to release a version of DLSS 4.5 that improves quality mode?

It’s unlikely that Nvidia will release a version of DLSS 4.5 that improves quality mode. The main focus of DLSS 4.5 is performance efficiency rather than image quality. Future improvements will likely be centered on optimizing performance in more aggressive presets like performance and ultra-performance.

Why is DLSS becoming more expensive to run as newer models are introduced?

DLSS becomes more computationally expensive as newer models are introduced because the improvements in visual quality and anti-aliasing require more GPU resources. As tensor cores become more capable, DLSS uses more of the frame time, which means older GPUs struggle more to handle the increased workload.

How do newer GPUs handle DLSS compared to older GPUs?

Newer GPUs, like the RTX 6090, can run DLSS with minimal impact on frame times, allowing for higher quality without significantly impacting performance. In contrast, older GPUs like the RTX 2080 Ti take longer to process DLSS, leading to greater performance trade-offs, especially in higher resolutions or more demanding workloads.

Should RTX 5090 users stick with DLSS 4.0 using quality mode, or switch to DLSS 4.5?

For RTX 5090 users, sticking with DLSS 4.0 in quality mode is likely a better option. DLSS 4.5, while providing benefits in performance modes, is not expected to offer a significant improvement in quality mode. The additional computational cost of DLSS 4.5 is not necessary for users with such powerful GPUs, where DLSS 4.0 can already deliver high-quality visuals.

What is the primary difference between the 'L' and 'M' DLSS presets?

'L' (ultra-performance) is focused on delivering performance gains at very low internal resolutions, often producing lower-quality visuals at minimal GPU cost. 'M' (presumably mid-performance) is targeted at providing a balance between quality and performance, offering better image quality at a moderate GPU cost compared to ultra-performance.

How does DLSS handle low internal resolutions, and what kind of results can users expect?

DLSS is capable of extracting significant image quality from low internal resolutions. For example, users have been able to run games like Control at 1440p with an internal resolution as low as 480p using DLSS ultra-performance, with surprisingly good visual results. This highlights DLSS’s ability to upscale low-resolution images effectively.

Why is Nvidia focusing on improving image quality in lower-performance modes like performance and ultra-performance?

Nvidia is focusing on improving image quality in lower-performance modes like performance and ultra-performance because these presets are more computationally demanding. As GPUs get more capable, improving these modes helps maintain high-quality visuals while keeping performance gains, especially in situations where users need to prioritize frame rates.

What role does the evolution of tensor core hardware play in the development of DLSS?

The evolution of tensor core hardware plays a crucial role in improving DLSS. As tensor cores become more capable, they enable better image reconstruction and processing, which leads to improvements in DLSS quality. However, this also means that DLSS becomes more computationally expensive over time, as it leverages the power of newer hardware to enhance visual fidelity.

How do the latest versions of DLSS compare to older anti-aliasing techniques like MSAA?

The latest versions of DLSS are more advanced than older anti-aliasing techniques like MSAA. While MSAA was effective in the past, it had significant performance costs, especially as game rendering became more complex. DLSS, on the other hand, uses deep learning to reconstruct images more efficiently, providing superior image quality with a lower computational cost, especially in modern GPUs.