GeForce 10 series
Nvidia's Pascal-based GPU series launched in May 2016.
The GeForce 10 series is a line of graphics processing units from Nvidia, built on the Pascal microarchitecture first announced in March 2014. This series replaced the GeForce 900 series and hit the market in May 2016. The initial models, the GTX 1080 and GTX 1070, were unveiled on May 6, 2016, with the 1080 launching on May 27 and the 1070 on June 10.
The Pascal architecture, named after Blaise Pascal, succeeded the Maxwell architecture. It used either 16 nm FinFET technology from TSMC or 14 nm FinFET from Samsung. Early chips were made exclusively on TSMC’s 16 nm process, but later chips like the GP107 and GP108 switched to Samsung’s 14 nm process.
Key features of the GP10x chips include CUDA Compute Capability 6.0 (GP100 only) or 6.1 (all others), DisplayPort 1.4 (without DSC), HDMI 2.0b, and fourth-generation Delta Color Compression. The PureVideo Feature Set H added hardware decoding for HEVC Main10 (10-bit), Main12 (12-bit), and VP9, which the previous GM200 and GM204 chips lacked. HDCP 2.2 support was included for 4K DRM content, a feature missing from Maxwell GM200 and GM204 but present in GM206. NVENC hardware encoding for HEVC Main10 (10-bit) was included, except on the GP108, which lacks NVENC entirely. Other additions were GPU Boost 3.0, Simultaneous Multi-Projection, HB SLI Bridge Technology, and a memory controller supporting both GDDR5X and GDDR5 (on GP102, GP104, and GP106).
The architecture introduced a dynamic load balancing scheduling system that adjusts GPU resources across tasks to keep the GPU fully occupied when work is available, enabling asynchronous compute in the driver. Instruction-level preemption was implemented: for graphics tasks, the driver uses pixel-level preemption (lower overhead), while compute tasks can use thread-level or instruction-level preemption (more expensive but necessary for long-running compute jobs). Triple buffering was added at the driver level, called Fast Sync, which maintains three frame buffers per monitor. This sends the most recently completed frame to the display, eliminating the initial delay of double buffering with vsync and preventing tearing. The trade-offs are higher memory usage and potential wasted power if multiple frames are drawn between display refreshes. This feature was backported to Maxwell GPUs in driver version 372.70.
Quick Facts
- Created
- May 17, 2016
- Manufacturer
- TSMC / Samsung
- Designfirm
- Nvidia
- Codename
- GP10x
- Architecture
- Pascal
- Model
- GeForce GTX series
- Process
- TSMC 16 nm (FinFET) · Samsung 14 nm (FinFET)
- Entry
- GeForce GT 1010 · GeForce GT 1030
- Midrange
- GeForce GTX 1050 · GeForce GTX 1050 Ti · GeForce GTX 1060
Facts from the source article.
Lore & Background
The GeForce 10 series is built on the Pascal microarchitecture, named after Blaise Pascal, which succeeded the Maxwell microarchitecture. The architecture incorporates either 16 nm FinFET (TSMC) or 14 nm FinFET (Samsung) technologies; initially chips were produced only in TSMC's 16 nm process, but later chips (GP107, GP108) were made with Samsung's 14 nm process. New features include CUDA Compute Capability 6.0 (GP100 only) and 6.1 (other GP10x chips), DisplayPort 1.4 (without DSC), HDMI 2.0b, fourth generation Delta Color Compression, PureVideo Feature Set H hardware video decoding (including HEVC Main10, Main12, and VP9), HDCP 2.2 support, NVENC HEVC Main10 10-bit hardware encoding (except GP108), GPU Boost 3.0, Simultaneous Multi-Projection, HB SLI Bridge Technology, a new memory controller with GDDR5X and GDDR5 support (GP102, GP104, GP106), dynamic load balancing scheduling, instruction-level preemption, and driver-level triple buffering called Fast Sync. The Pascal GP100 GPU features four High Bandwidth Memory stacks, allowing 16 GB HBM2, 16 nm technology, Unified Memory, and NVLink.
Reader's Guide
The GeForce 10 series marked a significant shift in notebook GPU design: notebook GPUs (GTX 1060–1080) had specifications close to their desktop counterparts, and the GTX 1050/1050 Ti even exceeded desktop specs, removing the 'M' suffix and enabling user overclocking. This was made possible by lower TDP ratings and improved thermal designs. The Max-Q Design, launched on June 27, 2017, offered lower-TDP, quieter variations for ultra-thin gaming systems. The GT series for notebooks was replaced by the MX series; only the MX150 was based on Pascal's GP108 die. Due to production problems surrounding RTX 30-series cards, a global semiconductor shortage, and increased cryptocurrency mining demand, the GTX 1050 Ti was brought back into production in 2021. Nvidia also quietly released the GeForce GT 1010 in January 2021. Driver support: 32-bit driver releases ended after version 391.35 in March 2018; Windows 7 and 8.1 driver support transitioned to legacy status in September 2024; the GeForce 10 series is the last Nvidia GPU generation to support 32-bit operating systems and the last to officially support Windows 8 and 8.1. In December 2025, Nvidia discontinued Game Ready Driver support for Maxwell, Pascal, and Volta architectures (including the GeForce 10 series), with security updates continuing until October 2028.
Did You Know?
- The GTX 1050 Ti was brought back into production in 2021 due to the global semiconductor shortage and cryptocurrency mining demand.
- Notebook GPUs in this series removed the 'M' suffix and offered performance close to desktop equivalents.
- The Pascal GP100 GPU features four High Bandwidth Memory stacks for up to 16 GB HBM2.
More in Nvidia and AMD Graphics Processors 1-24
Spotted an error? Know more?
Reader corrections go straight into our review queue. Suggest an edit · How this site is sourced