News
IndustryTSMC’s grip on advanced chipmaking will face competitionIndustryCadence joins European Chip Design Platform to support semiconductor startups and SMEsMaking chipsMarc Hijink on ASML’s Culture, Supply Chain, and GeopoliticsProcessorsIntel's post-Nova Lake CPUs take shape as Razor Lake desktop and laptop variants surfaceAI chipsHow Google’s New AI Chip Software Helps It Battle NvidiaIndiaMore IC Packaging Plants Emerge In IndiaIndustryUS Tightens China AI Chip Access as Nvidia, TSMC and Semiconductor Stocks SlideDesignAxelera AI moves into HPC with TitaniaMaking chipsChinese envoy says chip equipment export curbs threaten EU-China trade thawIndiaSEMICON India 2026 Day 2 Sees 25 MoUs, Strategic Collaborations Across Semiconductor Value ChainIndustryTSMC’s grip on advanced chipmaking will face competitionIndustryCadence joins European Chip Design Platform to support semiconductor startups and SMEsMaking chipsMarc Hijink on ASML’s Culture, Supply Chain, and GeopoliticsProcessorsIntel's post-Nova Lake CPUs take shape as Razor Lake desktop and laptop variants surfaceAI chipsHow Google’s New AI Chip Software Helps It Battle NvidiaIndiaMore IC Packaging Plants Emerge In IndiaIndustryUS Tightens China AI Chip Access as Nvidia, TSMC and Semiconductor Stocks SlideDesignAxelera AI moves into HPC with TitaniaMaking chipsChinese envoy says chip equipment export curbs threaten EU-China trade thawIndiaSEMICON India 2026 Day 2 Sees 25 MoUs, Strategic Collaborations Across Semiconductor Value Chain
EveryChip
GeForce GTX 1080 Ti (GP102)

Pascal

Nvidia GPU microarchitecture for gaming and data centres

Photo: Fritzchens Fritz, CC0, via Wikimedia Commons

Pascal is a graphics processing unit design by Nvidia, launched in 2016. It succeeded Maxwell and introduced high-bandwidth memory support and improved floating-point performance for deep learning tasks.

Share on WhatsApp

  • Launched in April 2016 with Tesla P100
  • Uses TSMC 16 nm or Samsung 14 nm process
  • Supports HBM2 and GDDR5X memory types
  • Introduces NVLink for high-speed GPU interconnect
  • Succeeded Maxwell and preceded Volta architectures

What it is

Pascal is a microarchitecture for graphics processors developed by Nvidia. It serves as the successor to the Maxwell design and was first introduced in April 2016. The architecture powers the GeForce 10 series for consumers and Tesla P series for servers. It supports advanced memory standards and improves floating-point calculation speeds for scientific and machine learning workloads.

How it came about

Nvidia announced Pascal in March 2014 as the next step after Maxwell. The first product using this design was the Tesla P100 accelerator, released on 5 April 2016. Consumer cards followed shortly, with the GTX 1080 launching in May 2016. The name honours Blaise Pascal, a French mathematician from the seventeenth century.

Inside the chip

The chip divides into Graphics Processor Clusters, each containing Streaming Multiprocessors. These units hold CUDA cores for parallel calculations. GP100 variants feature 64 cores per multiprocessor, while GP104 designs use 128. Pascal introduces unified memory access for CPUs and GPUs via a Page Migration Engine. It also supports NVLink for fast data transfer between processors and enables half-precision math at double the rate of single-precision operations on specific models.

How it was made

Manufacturers produced Pascal chips using two different processes. TSMC fabricated many units on its 16 nanometre FinFET node. Samsung also produced variants using its 14 nanometre FinFET technology. This dual sourcing allowed Nvidia to meet demand across different product lines, from high-end data centre accelerators to mainstream gaming cards.

Pascal
A drawing made by EveryChip from what this page says. It is a diagram, not a picture of the real chip.

Where it went

This architecture appears in desktop gaming cards like the GTX 1070 and GTX 1080. Professional users employ Quadro P series workstations for visualisation. Data centres rely on Tesla P40 and P100 accelerators for deep learning training and high-performance computing. The design handles video decoding for modern formats and supports virtual reality displays through high-speed output interfaces.

Versions and family

Several silicon implementations exist. GP100 targets professional compute with HBM2 memory. GP102 powers the Titan Xp and GTX 1080 Ti. GP104 serves the GTX 1080 and 1070 models. Smaller dies include GP106 for the GTX 1060, GP107 for the GTX 1050, and GP108 for entry-level GT cards. Each variant adjusts core counts and memory interfaces to fit specific market segments.

Why it matters

Pascal established new performance baselines for consumer graphics in 2016. It brought high-bandwidth memory to mainstream professional cards and improved efficiency for gaming. The introduction of NVLink changed how multiple GPUs communicate in server environments. Later software updates enabled ray tracing features on these older cards, extending their useful life beyond the initial launch period.

Questions people ask about Pascal

What is the Pascal microarchitecture?

Pascal is a GPU design by Nvidia released in 2016. It sits between Maxwell and Volta in the company's timeline. It focuses on improving performance for gaming and deep learning through better memory bandwidth and calculation efficiency.

Which chips use the Pascal design?

Consumer cards include the GeForce GTX 10 series, such as the GTX 1080 and 1070. Professional products include Quadro P series and Tesla P40 or P100 accelerators. The underlying silicon varies from GP100 to GP108 depending on the model.

What manufacturing process was used for Pascal?

Nvidia used two foundries for production. TSMC manufactured chips on a 16 nanometre FinFET process. Samsung produced other variants using a 14 nanometre FinFET node. This allowed for flexible supply chain management during the product lifecycle.

Does Pascal support ray tracing?

Hardware ray tracing was not built into Pascal. However, Nvidia enabled a software implementation of DirectX Raytracing in 2019 for certain cards. This feature was initially reserved for newer Turing architectures but became available on Pascal via driver updates.

Good reading and sources

Where this page comes from

The facts on this page come from the Wikipedia article "Pascal (microarchitecture)", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryChip rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.

It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

Photo credits on this page

The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.

The weekly digest

Every Monday: the chip of the day, new pages on EveryChip, and the conferences and anniversaries of the week ahead.