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
GP5

GP5

Analog co-processor for belief propagation and machine learning tasks

The GP5 is a specialised co-processor designed to speed up discrete belief propagation on factor graphs. Developed under the DARPA Analog Logic programme, it uses an exotic architecture distinct from GPUs or DSPs. Analog Devices acquired the technology in 2011 via its purchase of Lyric Semiconductor.

Share on WhatsApp

  • Accelerates discrete belief propagation on factor graphs
  • Developed under the DARPA Analog Logic programme
  • Acquired by Analog Devices in 2011 via Lyric Semiconductor
  • Uses heterogeneous compute units with deep pipelining
  • Predated the Google Tensor Processing Unit by several years

What it is

The GP5 functions as a hardware accelerator for specific mathematical operations used in machine learning. It targets discrete belief propagation on factor graphs and large-scale tensor products. Unlike general-purpose processors, this chip acts as a companion to a main controller, such as an x86 CPU or an ARM core. Its design stems from the DARPA Analog Logic initiative, aiming to handle complex probabilistic calculations more efficiently than traditional digital logic allows.

How it came about

This component emerged from research funded by the Defense Advanced Research Projects Agency under its Analog Logic programme. Lyric Semiconductor developed the technology as a culmination of this effort. The design predated similar commercial efforts, including the Google Tensor Processing Unit, by several years. In 2011, Analog Devices purchased Lyric Semiconductor, thereby acquiring the intellectual property rights to the GP5 architecture and related technologies.

Inside the chip

The architecture avoids standard GPU or DSP structures, employing a unique mix of fine-grained and coarse-grained parallelism. It features deep pipelining and independent heterogeneous compute units that handle different parts of belief propagation updates simultaneously. Performance depends heavily on the specific workload structure. Benchmarks suggest it offers significant speed and energy advantages over modern CPUs, GPUs, and ARM processors for tasks like protein folding, error correction decoding, and stereo vision.

How it was made

Specific details regarding the manufacturing process node, transistor count, or physical packaging are not provided in the available records. The chip was developed by Lyric Semiconductor before its acquisition by Analog Devices. The focus of its creation was on the logical architecture and algorithmic efficiency rather than publicised fabrication metrics. It represents an application of analog logic principles to digital-style computational problems.

Where it went

Engineers tested the chip on various inference and machine learning workloads. These applications included reducing signal noise, processing stereo vision data, decoding turbo errors, and folding protein side-chains. It operates alongside a host processor, offloading these intensive mathematical tasks to improve overall system efficiency. Its primary role is accelerating probabilistic graphical models used in advanced computational scenarios.

Versions and family

No specific variants, siblings, or direct successors are listed in the records. The GP5 stands as a distinct product of the Lyric Semiconductor portfolio before its integration into Analog Devices. It is noted for anticipating later industry trends in specialised AI accelerators, but no direct lineage of subsequent chips is detailed in the provided information.

Why it matters

The GP5 demonstrated that analog-inspired architectures could outperform conventional digital processors in specific niche tasks. By achieving higher energy efficiency and speed for belief propagation, it highlighted potential pathways for future AI hardware. Its early development relative to major industry players like Google suggests it was a pioneer in dedicated tensor operation acceleration, influencing the broader understanding of non-standard computing paradigms.

Questions people ask about GP5

What is the primary function of the GP5 chip?

The GP5 accelerates discrete belief propagation on factor graphs and large-scale tensor product operations. It serves as a co-processor for machine learning tasks, working alongside a main CPU or controller to handle specific probabilistic calculations.

Who originally developed the GP5 architecture?

Lyric Semiconductor designed the GP5 as part of the DARPA Analog Logic programme. Analog Devices later acquired the intellectual property when it bought Lyric Semiconductor in 2011.

How does the GP5 architecture differ from a GPU?

The GP5 uses an exotic architecture that differs from both GPUs and DSPs. It leverages massive fine-grained and coarse-grained parallelism with independent heterogeneous compute units, rather than following standard graphics processing designs.

Good reading and sources

Where this page comes from

The facts on this page come from the Wikipedia article "GP5 chip", 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.