Early RTX 5080 Benchmarks Show an 8.3% Edge Over the RTX 4080 SUPER Without Upscaling

NVIDIA is set to unveil its next-generation Blackwell graphics cards on January 30, but early performance leaks for the RTX 5080 have already ignited heated discussions among gamers. According to insider reports, the base model shows only modest gains, while the flagship RTX 5090 promises a dramatic leap in power.

Reddit users recently uncovered a deleted review of the RTX 5080 by an independent tester, showcasing benchmarks across 11 games at 1440p with native rendering. The card reportedly outpaces the RTX 4080 SUPER by 8.3% while drawing 360W of power—a 4% increase over its predecessor. However, the gap widens in DLSS or ray-traced scenarios, thanks to upgraded RT cores and faster GDDR7 memory bandwidth.

Despite sharing a similar CUDA core count with the RTX 4080 SUPER (10,752 vs. 10,240), the RTX 5080 features a revamped 256-bit memory interface and clock speeds up to 2.62 GHz. Even so, it still trails the RTX 4090 by 15% in raw performance. Meanwhile, insider claims suggest the RTX 5090 delivers 50% higher performance than the RTX 4080 SUPER and outpaces the RTX 5080 by 38%, albeit at a hefty 434W power draw.

Official reviews of the RTX 5080 and RTX 5090 are expected today, but one trend is clear: NVIDIA appears to be prioritizing niche capabilities—like ray tracing and upscaling—over substantial generational leaps in raw performance. For gamers hoping for a revolution, the real test will be how effectively developers optimize games for the new RT cores and whether DLSS 4.0 can offset the modest native gains.