RTX 5090 tops RTX 5080 tests
- PCBench’s latest GeForce head-to-head puts Nvidia’s RTX 5090 ahead of the RTX 5080 across 73 recent games and several synthetic benchmarks. - The clearest gap shows up at 4K and ray tracing: other test suites have the 5090 running roughly 30% to nearly 69% faster. - That speed comes with a brutal tradeoff — $1,999 MSRP, 575 W power, and weak value next to the $999 RTX 5080.
Nvidia’s top Blackwell gaming card is doing exactly what a halo GPU is supposed to do — win the benchmark charts. PCBench’s fresh RTX 5090 vs. RTX 5080 roundup shows the 5090 on top across 73 recent games plus synthetic tests, which is the simple version of the story. But the interesting part isn’t that Nvidia’s most expensive card is faster. It’s how much faster, where that lead actually shows up, and why a lot of people still won’t call it the better buy. (pcbench.net) ### What actually got tested? PCBench compared the GeForce RTX 5080 and RTX 5090 using a large game sample and synthetic runs, with spec sheets alongside the results. On paper, the 5090 is a monster: 32 GB of GDDR7, a 512-bit bus, 1,790 GB/s of bandwidth, and 21,760 CUDA cores. The 5080 is still(pcbench.net)e gap is why nobody expected a close race. (pcbench.net) ### So how big is the lead? In PCBench’s synthetic slice, the 5090 is about 47% ahead in 3DMark Time Spy and 41% ahead in Geekbench 6 OpenCL. In games, the gap moves around a lot more. At lower resolutions, CPU limits and lighter workloads can compress the difference. At 4K — especially with ray(pcbench.net)80 at 4K in its own testing, which lines up with the basic shape PCBench is showing. (pcbench.net) ### Why does 4K change the story? Because 4K is where raw GPU muscle matters most. Once you push ultra settings, heavy ray tracing, and big memory loads, the 5090’s extra bandwidth and VRAM start to matter in a way they often don’t at 1080p. Think of it like two sports cars on city streets versu(pcbench.net) straight. That is basically the 5090 at 4K. (pcbench.net) ### What about DLSS 4? This generation’s wildcard is DLSS 4 with Multi Frame Generation. Nvidia launched RTX 5090 and 5080 support with DLSS 4 and said more than 75 games and apps had Multi Frame Generation support on day one. That matters because some of the eye-popping frame-rate numbers peopl(pcbench.net)ring and AI frame generation. The 5090 still benefits most, but the software stack is part of the performance story now. (nvidia.com) ### Then why is the value argument so harsh? Price and power. Nvidia launched the RTX 5090 at $1,999 and the RTX 5080 at $999. The 5090 also carries a 575 W board power figure, versus 360 W for the 5080. So you are not paying double for a clean doubling of gaming performance. You are paying flagship tax for the las(nvidia.com)ache. (nvidianews.nvidia.com) ### Does that spill into laptops too? Absolutely — and the prices get silly fast. ASUS’s 2026 ROG Zephyrus Duo can be configured with an RTX 5090 Laptop GPU, and ASUS lists the line starting at $4,499.99 in the U.S. Geeknetic reviewed a higher-end configuration at €6,499.99. So the desktop “fast but hard to justify” debate gets even sharper once the 5090 badge moves into premium laptops. (rog.asus.com) ### Who is the 5090 really for? People chasing the absolute top end — 4K max settings, heavy ray tracing, creator workloads, AI experiments, or just no-compromise builds. For almost everyone else, the RTX 5080 is the saner card. It gives up the crown, but not in a way most players will feel enough to justify t(rog.asus.com)ins the race, but the 5080 makes the argument.
