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
Project Denver

Project Denver

Nvidia ARM CPU with binary translation

Project Denver is a central processing unit designed by Nvidia using the ARMv8-A instruction set. It employs dynamic binary translation to convert ARM code into internal VLIW instructions for high performance in mobile and server systems.

Share on WhatsApp

  • Uses dynamic binary translation to convert ARM code to VLIW
  • First shipped in the HTC Nexus 9 tablet in November 2014
  • Reserves 128 MiB of main memory for code translation cache
  • Carmel variant supports ARMv8.2-A with functional safety
  • Originated from Stexar Company acquisition in 2006

What it is

Project Denver is a central processing unit developed by Nvidia that implements the ARMv8-A 64-bit and 32-bit instruction sets. Unlike traditional designs, it uses a combination of simple hardware decoding and software-based binary translation. This dynamic recompilation layer runs below the operating system, storing optimised code sequences in a dedicated portion of main memory. The architecture is built as a very wide in-order superscalar pipeline, making it suitable for integration into systems on a chip alongside graphics and other processors.

How it came about

Nvidia revealed Project Denver at the 2011 Consumer Electronics Show. CEO Jen-Hsun Huang stated that hundreds of engineers had worked on the five-year development programme for three and a half years. The project originated at Stexar Company in Colorado, which Nvidia acquired in 2006. Stexar initially aimed to create an x86-compatible processor using binary translation similar to Transmeta. The team included engineers from Intel, AMD, HP, Sun and Transmeta. The design shifted to ARMv8-A after Nvidia could not secure an Intel patent license. The first consumer device, the HTC Nexus 9 tablet, launched in November 2014.

Inside the chip

The processor features a two-way decoder for ARM instructions and translates them on the fly into internal Very Long Instruction Word (VLIW) instructions. A hardware translator handles this process, with software emulation as a fallback. The translation can reorder instructions and remove those that do not affect the result. It executes up to seven micro-operations per clock cycle using translated VLIW instructions. Each core has 128 KiB of instruction cache and 64 KiB of data cache. Two cores share a 2 MiB L2 cache. The system reserves 128 MiB of main memory for the translation cache, which the operating system cannot access.

How it was made

The initial Denver cores were manufactured using the 28 nm process node by TSMC. These chips supported clock speeds up to 2.5 GHz. The improved Denver 2 design moved to a more efficient 16 nm node. The subsequent Carmel design, released in 2018, used a 12 nm structure size. Carmel supports ARMv8.2-A with functional safety features, dual execution, parity and error-correcting code protection. It also includes full Advanced SIMD and Vector Floating Point support.

Where it went

Denver cores are integrated into Nvidia Tegra systems on a chip for mobile computers, personal computers, servers and supercomputers. The dual-core Denver CPU paired with a Kepler GPU formed the Tegra K1, used in the Google Nexus 9 tablet. The Tegra X2 combined two Denver 2 cores with four Cortex-A57 cores in a heterogeneous multi-processor arrangement, paired with a Pascal GPU. The Tegra Xavier SoC features eight Carmel cores divided into four blocks, matched with a Volta GPU and special purpose accelerators.

Versions and family

The original Denver 1 appeared in the Tegra K1 (model T132). Denver 2, built on the 16 nm node, was used in the Tegra X2 (model T186). Carmel, based on ARMv8.2-A, succeeded these designs in the Tegra Xavier (model T194). Carmel offers ten-way superscalar execution and supports ARMv8.2-FP16. While the Tegra K1 name also applies to a quad-core variant, that version does not use Denver cores. Carmel configurations include four dual-core pairs sharing 4 MiB of L3 memory.

Why it matters

Project Denver demonstrated that dynamic binary translation could deliver high performance on ARM architectures. It enabled Nvidia to enter the 64-bit ARM server and mobile markets with custom silicon. The Nexus 9 was the first 64-bit Android device available to consumers. Third-party tests of the Carmel cores in the Jetson AGX development kit in September 2018 showed noticeable performance increases over predecessors. The architecture allowed Nvidia to integrate CPU, GPU and accelerators into single dies for diverse computing tasks.

Questions people ask about Project Denver

How does Project Denver execute ARM instructions?

It uses a hardware translator to convert ARM instructions into internal VLIW instructions on the fly. Software emulation acts as a fallback. The system stores optimised code sequences in a 128 MB cache within main memory, allowing it to reorder instructions for better performance.

Which devices used the Denver CPU cores?

The first consumer device was the Google Nexus 9 tablet, which used the Tegra K1 SoC with dual Denver cores. Later versions appeared in the Tegra X2 and Tegra Xavier systems on a chip, targeting mobile devices, servers and supercomputers.

What is the difference between Denver and Carmel?

Carmel is an improved design released in 2018 based on ARMv8.2-A. It uses a 12 nm process compared to 28 nm or 16 nm for earlier Denver versions. Carmel adds functional safety, dual execution, and support for ARMv8.2-FP16 instructions.

Good reading and sources

Where this page comes from

The facts on this page come from the Wikipedia article "Project Denver", 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.