XMOS
XMOS is a fabless semiconductor company that develops voice solutions, audio products, and multicore microcontrollers capable of concurrently executing real-time tasks, DSP, edge AI processing, and control flow. XMOS microcontrollers are distinguished by their deterministic behavior.
Company history
XMOS was founded in July 2005 by Ali Dixon, James Foster, Noel Hurley, David May, and Hitesh Mehta. It received seed funding from the University of Bristol enterprise fund, and Wyvern seed fund.The name XMOS is a loose reference to Inmos. Some concepts found in XMOS technology are part of the Transputer legacy.
In the autumn of 2006, XMOS secured funding from Amadeus Capital Partners, DFJ Esprit, and Foundation Capital. It also has strategic investors Robert Bosch Venture Capital GmbH, Huawei Technologies, and Xilinx Inc. In September 2017, XMOS secured
$15M in an investment round lead by Infineon.
XMOS's processor technology is general-purpose and has been exploited in a range of different markets, including voice, microphone arrays, audio, LED tiles, communications, and robotics.
This enables third parties to establish products and businesses based around the technology.
In July 2017, XMOS acquired SETEM, a company that specialises in audio algorithms for source separation.
Products
XMOS has developed families of silicon devices and software based on xCORE technology:- xcore.ai - Announced in February 2020, xcore.ai devices build on the xCORE-200 series, adding hardware 32-bit floating-point capability and 256-bit vector processing. These processors are aimed at embedded and IoT devices utilizing AI acceleration in SoC-like designs. Interfaces for external LPDDR memory and MIPI camera or sensor input are available, along with high-speed USB.
- xCORE VocalFusion - Launched in June 2017. xCORE VocalFusion devices combine far-field voice capture technologies and acoustic digital signal processing in a single device that can also include support for Sensory, Inc. TrulyHandsfree Voice Control. The VocalFusion 4-Mic Kit for the Alexa Voice Service and VocalFusion Stereo Dev Kit for Amazon AVS have been qualified by Amazon.
- xCORE-VOICE - Launched in April 2016. xCORE-VOICE processors are a combination of one or more xCORE-200 processors with software that enables capture of multiple microphone signals, and aggregate those into a single signal.
- xCORE-AUDIO - Released in April 2016. xCORE-AUDIO are audio-specific processors built on xCORE-200 technology, aimed at high-resolution consumer audio and multichannel professional audio applications.
- xCORE-200 - Released in March 2015. xCORE-200 devices can execute dual-issue code at 500 MHz, have up to four tiles, with each tile executing up to eight concurrent tasks. The devices are packaged with options for embedded flash, embedded USB PHY, and an embedded RGMII interface.
- XCore XS1-L1 - Released from 2008. The first family of multicore microcontrollers, with up to two tiles executing at up to 500 MHz. There are options for embedded USB, ADCs, DC-DC converters, or an integrated ARM Cortex-M3.
Voice processing
XMOS launched its first microphone array for voice capture in 2016 for far-field voice capture applications.In November 2016 XMOS partnered with Sensory, Inc. to deliver TrulyHandsfree voice control technology on XMOS voice processing platforms.
In March 2017, Infineon Technologies and XMOS demonstrated a combination of radar and silicon microphone sensors from Infineon and audio processor from XMOS at GSMA Mobile World Congress in Barcelona.
The XVF3000 family of far-field voice processors announced in June 2017 includes voice pre-processing DSP for adaptive beamforming, full-duplex acoustic echo cancellation with barge-in, noise suppression, automatic gain control.
In October 2017 the XMOS VocalFusion 4-Mic Dev Kit for Amazon AVS was qualified by Amazon.
In February 2018 the VocalFusion Stereo Dev Kit for Amazon AVS was announced at Mobile World Congress 2018, and qualified by Amazon for the Alexa Voice Service.
Digital audio
XMOS multicore microcontrollers are used extensively by multinational companies such as Audio Partnership, Cambridge Audio, FiiO Electronics Technology, Meridian Audio, Native Instruments, Oppo Digital, Sennheiser and Sony to implement USB Audio 2.0 interfaces in their products.In December 2014, AVnu Alliance, the industry consortium driving open standards-based deterministic networking through certification, announced XMOS as the first available AVnu-certified Audio Video Bridging and Time-Sensitive Networking audio endpoint reference platform.
xCORE multicore microcontrollers
xCORE multicore microcontrollers comprise one or more processor tiles connected by a high-speed switch. Each processor tile is a conventional RISC processor that can execute up to eight tasks concurrently. Tasks can communicate with each other over channels, or using memory.The xCORE architecture delivers, in hardware, many of the elements that are usually seen in a real-time operating system. This includes the task scheduler, timers, I/O operations, and channel communication. By eliminating sources of timing uncertainty, xCORE can provide deterministic and predictable performance for many applications. A task can typically respond in nanoseconds to events such as external I/O or timers. This makes it possible to program xCORE devices to perform hard real-time tasks that would otherwise require dedicated hardware.
xCORE devices have been used in a range of different markets, including voice processing, USB Audio, AVB and Time-Sensitive Networking, industrial communications, and robotics.
xCORE devices can be programmed using C, C++, xC or native assembler. To help programmers access the real-time hardware features of xCORE devices, some multicore language extensions for C have been added. These extensions form a programming language called xC which contains features for task-based parallelism and communication, accurate timing and I/O, and safe memory management.
A tool-chain, xTIMEcomposer, comes with LLVM-based compilers for C, C++ and xC, cycle-accurate simulator, symbolic debugger, runtime instrumentation and trace libraries and a static code timing analyzer. All of the components are aware of the real-time multicore nature of the programs, giving a fully integrated approach.