Impressive when it first came out, the GeForce GTX 1080 is now something of a museum piece.
Just over ten years ago - on May 6, 2016, to be exact - NVIDIA launched a generation that seemed at first glance to be the logical continuation of years of evolution. The so-called "Pascal" generation, heralded by the GeForce GTX 1070 and GTX 1080, was indeed marketed by NVIDIA, which took advantage of the opportunity to push the envelope a little further in relation to AMD, even though the previous generation - the "Maxwell" generation launched two years earlier - had already enabled GeForce to widen the gap with Radeon.
When NVIDIA detailed the GeForce GTX 1080... with a higher price tag for its Founders Edition: times have changed! NVIDIA
Engraved in 16nm FinFET by TSMC, the GeForce GTX 1080's GPU GP104 featured 2,560 CUDA cores. It operated at a base frequency of 1607 MHz and was capable of a boost up to 1733 MHz. On the video memory side, it was backed up by 8 GB GDDR5X running at 10 Gb/s on a 256-bit bus, for a bandwidth of 320 GB/s. Beyond these figures, the GeForce RTX 1080 marked the end of an era. Indeed, after the release of this "Pascal" generation, NVIDIA began to focus on new technologies, which led to the revolution we know today in the way we approach video game development.
At the time of the GeForce GTX 1080, NVIDIA wasn't talking about ray tracing, Deep Learning Super Sampling (DLSS) or Tensor cores. These technologies were so far removed from us that it wasn't even necessary to talk about "rasterization" for what gamers felt was the only way to render in a video game. Today, DLSS is on all the most demanding games - just like AMD and Intel's FSR and XeSS - and, between upscaling and multi-frame generation, artificial intelligence is responsible for "thinking" practically seven out of every eight pixels displayed in today's games! NVIDIA cards are now "RTX" models, and these changes have taken place in less than ten years... What will it be like in 2036?
