NVIDIA GTX 10 series at 10, from Pascal to RTX
NVIDIA announced the GeForce GTX 1080 on May 6, 2016, marking the start of the GeForce GTX 10 series for gamers. The card was based on the Pascal architecture and launched at $599, with a $699 Founders Edition. Retail availability followed on May 27, 2016, while the GeForce GTX 1070 arrived on June 10.
Pascal: no Ray Tracing and no DLSS, just pure performance
Pascal was one of NVIDIA’s last GeForce architectures sold mostly on raster gaming performance, power efficiency, and VR. This was before RTX branding, RT cores, Tensor cores, DLSS, and AI features became part of the gaming GPU stack. The GTX 10 series still introduced a large process jump to 16nm FinFET and support for features such as DisplayPort 1.4, HDMI 2.0b, GPU Boost 3.0, and faster memory.
Source: NVIDIA
The GeForce GTX 1080 used the GP104 GPU with 2560 CUDA cores, a 1607 MHz base clock, a 1733 MHz boost clock, and 8GB of GDDR5X memory on a 256-bit bus. NVIDIA listed memory speed at 10Gbps, giving the card 320GB/s of bandwidth. The GeForce GTX 1070 used the same GP104 silicon in a cut-down configuration with 1920 CUDA cores, 8GB of GDDR5 memory, a 256-bit bus, 256GB/s of bandwidth, and a 150W TDP.
The GTX 10 series also became tied to the cryptocurrency mining wave that followed in 2017 and 2018. Cards such as the GTX 1060, GTX 1070, and GTX 1080 Ti became harder to find at normal retail prices. This changed how many gamers remember Pascal, since the cards launched at relatively reasonable MSRPs but later became part of one of the first large GPU supply and pricing issues of the modern era.
The family that gave us the GTX 1080 Ti
NVIDIA later added the GeForce GTX 1080 Ti in 2017. It used the larger GP102 GPU with 3584 CUDA cores, 11GB of GDDR5X memory, an 11Gbps memory speed, and a 352-bit bus. At $699, it delivered performance close to the Pascal TITAN cards at a much lower launch price. For many PC gamers, the GTX 1080 Ti remains one of the best high-end GPU releases of the past decade in price-to-performance terms.
Source: NVIDIA
