When you buy a high-end GeForce RTX GPU, you expect to take advantage of every feature Nvidia loves to boast about. That's exactly what I thought when I picked up my first RTX card, the RTX 3090, back in 2020. Fast forward to 2025, and even after upgrading to an RTX 4090, I still turn off RTX features in most of the games I play. You'd think that with such a powerful card, I'd be chasing maxed-out graphics settings in every title, but to me, the trade-offs just aren't worth it.

I'm not saying features like ray tracing and DLSS aren't worth enabling at all, but I've found that they shine only in very specific scenarios. For the kinds of games I play and the way I like to enjoy them, the drawbacks often outweigh the wow factor. DLSS upscaling, for example, introduces artifacts in fast-paced scenes, whereas frame generation adds latency, which is the last thing you'd want as a competitive gamer. I'll admit that avoiding RTX features with a high-end GPU might seem counterintuitive, but once you go through these reasons, you'll see why my choice is justified.

The FPS hit outweighs the visual upgrade

Most games don't implement ray tracing well enough to justify the cost

From my experience, I have to sacrifice about 30-50% of the frame rate whenever I turn on ray tracing in most games that support it. I wouldn't mind enabling ray tracing if I averaged over 150FPS in a game without it. At that point, I know that even with ray tracing turned on, my average FPS would still be well above 60 for a fairly smooth experience. But the thing is, the visual upgrade doesn't really justify the performance hit. Sure, the games look better, but not enough to make me want to deal with dropped frames and microstutters.

Unfortunately, most games that support ray tracing don't implement the feature well enough to deliver a truly transformative experience. Many rely on ray-traced reflections, shadows, or ambient occlusion, which add subtle improvements you'd struggle to notice unless you paused and compared frames side by side. To make ray tracing feel worthwhile, a game needs to support path tracing or at least ray-traced global illumination. But in those cases, you are forced to rely on DLSS upscaling and frame generation to keep the game playable because your FPS tanks by more than 50%, and they both have their own drawbacks.

DLSS upscaling isn't flawless

DLSS 4 is a huge improvement, but native rendering still has the edge

There's no doubt that DLSS 4 has further narrowed the gap between upscaling and native rendering. Nvidia's new transformer model for DLSS Super Resolution does a much better job of reconstructing detail and maintaining motion stability that you'd struggle to notice the difference between upscaled and natively rendered footage in slow-paced single-player games. However, when you play fast-paced games, especially at 1440p or 1080p resolution, you'll quickly realize why it still falls short.

For starters, when you're panning the camera around in fast-paced titles, you'll notice that the textures and fine details don't hold up as well as they do at native resolution. Thin objects like fences and foliage can shimmer during quick movements, especially at lower resolutions like 1440p. And although DLSS 4 significantly reduces ghosting and blur compared to DLSS 3.5, I wouldn't say they're completely gone. If you sit pretty close to the monitor as I often do while playing first-person shooters, those subtle softness and shimmering issues become harder to overlook.

Frame generation comes at a cost

DLSS frame generation is far from ideal for competitive gaming

DLSS frame generation surely has its place, but it's not the magic solution for the kind of games that I play. While it does undoubtedly make AAA single-player games look smoother, the trade-offs become obvious in fast-paced competitive titles. That's mainly because the FPS boost you get from frame generation doesn't reduce input latency, since your GPU still renders the same number of frames natively. In fact, it actually adds latency because the GPU needs time to analyze the last rendered frame and motion data before predicting the next one using AI.

Anyone who plays competitive games knows just how crucial responsiveness is; even a few additional milliseconds of latency can be the difference between killing your opponent and getting killed. Frame generation improves the perceived smoothness of animations, but it doesn’t actually make the game feel smoother to play, since your inputs are still tied to the native render rate. The lower the base render rate, the more obvious this problem becomes. On top of that, it can introduce distracting artifacts like ghost frames, animation glitches, and other visual distortions during chaotic moments like gunfights, which only makes me avoid it altogether.

Rasterization still gives me the experience I actually want

I do experiment with RTX features from time to time, but only when a game makes it truly worthwhile, like Cyberpunk 2077's RT Overdrive setting, for example. Considering we rarely get games that properly implement ray tracing, I'd rather leave RTX features off and enjoy the high frame rates, low latency, and consistent smoothness that rasterization provides. Sure, you could argue that I could leave ray tracing off and benefit from DLSS upscaling and frame generation, but the latency hit and occasional artifacts in the fast-paced games I mostly play just don't make the experience worth it for me. As much as I love chasing high frame rates, I've learned that responsiveness is what really defines how a game feels. Until Nvidia improves high-end RTX GPUs to the point where ray tracing is feasible at high frame rates without relying on DLSS, I'm not ready to give up rasterization just yet.