The RTX 5080 launches tomorrow, so I wanted to get real numbers on the table—because after testing the 5090, the vibe for this generation was already… not great. On paper, the 5080 is a $1,000 card with 16GB of VRAM (now GDDR7 for more bandwidth), a slightly smaller process, and a few more shaders. In reality, across the 12 games I tested, the generational uplift versus the RTX 4080 averages about 14%, and versus the RTX 4080 Super (which was also $1,000) it’s more like 9%. In some games it’s basically identical, which is genuinely wild for a new “80-class” GPU. Power and thermals are the one bright spot: despite a 360W rated TDP, I was seeing around 273W average, and the Founders Edition cooler (same core design as the 5090 as far as I can tell) held about 65°C after 30 minutes at full load on an open bench. The problem is that the low power draw makes the 5080 look even weaker—I’d rather it ran closer to 320W like the 4080 Super and actually delivered more performance. If you’re coming from a 3080 or below, sure, it can be a meaningful upgrade (around 70% on average), but compared to what “80-tier” cards used to be—beating the previous flagship—this one doesn’t even threaten the 4090. It feels more like a 5070. I also talk about DLSS 4 Multi Frame Gen: I’m not fully anti-frame-gen, but it depends massively on your starting FPS. On the 5080, the lower base framerate made artifacting and latency worse compared to the 5090. In the real world, if a 4080 Super is cheaper or a used 4090 is only a bit more, I’d rather go that route than blindly throw money at the 5080.