When I first upgraded from the RTX 3090 to the RTX 4090 in 2022, I immediately noticed I wasn't getting anywhere close to the 60% performance uplift that various benchmarks suggested. Considering my Ryzen 9 5900X was one of the best CPUs available when the RTX 3090 launched, I didn't bother upgrading my CPU, especially since Zen 4 chips had only started rolling out a few weeks before the RTX 4090's release. But the more I gamed, the more I realized that my CPU just couldn't keep up with the card.

So how did I find this out, you ask? Well, the moment I started monitoring performance, the clues were suddenly everywhere. For starters, my GPU usage was unusually low, rarely exceeding 80% at 1440p, and sometimes dropping close to 60% in competitive titles. Frame rates barely improved despite lowering graphics settings, and 1% lows were all over the place. What I initially dismissed as driver quirks or game optimization issues turned out to be a textbook CPU bottleneck.

GPU usage simply didn't make sense

My RTX 4090 remained underutilized, no matter what game I played at 1440p

At 1440p, I didn't expect the RTX 4090 to be pinned at 95-99% utilization unless I was playing a graphically demanding AAA title like Cyberpunk 2077 with RT Overdrive enabled. But the problem is, my GPU usage rarely exceeded 80% at this resolution, even while playing AAA titles without ray tracing. And in competitive titles like Valorant and Counter-Strike 2, the usage would even dip to around 60% at times. That kind of behavior was something I had never encountered before, and it became clear my GPU wasn't performing at its full potential.

When you notice that your GPU temps and boost clocks look perfectly normal, yet usage remains unusually low, it makes you wonder what's holding performance back. Pointing fingers at my CPU made the most sense, especially once I started thinking about how frames are actually delivered. After all, my GPU can only render as fast as my CPU feeds it data, and consistent underutilization signals that the card is waiting rather than working. That, however, was only one of the clues suggesting my Ryzen 9 5900X was holding my RTX 4090 back.

Lowering graphics settings barely improved FPS

Giving my RTX 4090 more headroom wasn't helping it stretch its legs

Normally, when you lower graphics settings, you expect at least some measurable improvement in frame rates. When your graphics card doesn't have to render high-resolution textures, shadows, and complex lighting, it should be able to use that extra headroom to push FPS higher. That is one of the most basic indicators of a GPU-bound scenario, which is why lowering settings is often my first step when performance falls short of expectations. But when your CPU is the one limiting its performance, doing this will barely move the needle.

That’s exactly what I kept running into with my RTX 4090 at 1440p. But since my GPU already had plenty of headroom, given how low its usage was, there was barely any FPS difference between playing at Ultra and Low settings. If anything, the gains were limited to the single digits, which I suspect mostly came from settings that reduced CPU load rather than GPU load. In fact, I could even play on my 4K monitor at similar frame rates, despite the higher rendering demand. That kind of performance scaling behavior made it very clear that my 5900X was the bottleneck.

My 1% and 0.1% lows were hard to overlook

Frame time consistency, not average FPS, was the biggest problem

Honestly, I'd struggle to tell the difference between 200FPS and 300FPS unless I'm actively monitoring frame rates using MSI Afterburner. When you're playing competitive titles on a 360Hz OLED monitor, consistency matters more than average frame rates. I wouldn't have started monitoring performance if it weren't for the subtle FPS dips that kept annoying me during gaming sessions. Those momentary drops were far more noticeable than any difference in average frame rates.

When I finally checked the 1% and 0.1% lows in Afterburner, the underlying issue became much easier to understand. Those lows clearly indicated why gameplay felt inconsistent despite triple-digit average frame rates. A split-second dip from 200FPS to 100FPS may not seem alarming at first glance, but on a fast OLED panel, it's impossible to ignore. Considering I hadn't experienced these major dips when I had the RTX 3090, I couldn't come to terms with the fact that my far more powerful RTX 4090 somehow delivered a less consistent gaming experience — unless my CPU simply couldn't keep up.

Upgrading to the 5800X3D changed everything

When I finally came to terms with the fact that I was never going to get the most out of my RTX 4090 until I upgraded my CPU, I caved in and picked up a used 5800X3D for below its launch MSRP. And that alone felt like another GPU upgrade because my GPU usage immediately jumped, frame rates scaled the way they should have from the start, and those frustrating dips that once defined the experience largely disappeared. Even though it's just an 8-core CPU compared to my 12-core Ryzen 9 5900X, its massive L3 cache proved far more valuable for gaming, dramatically improving frame time consistency and allowing my RTX 4090 to finally stretch its legs.