CALIBRE

WittyBox™ is a new series of general-purpose avionic computers for aerospace and mission systems. It is based on a modular architecture with pre-certified aerospace standards, allowing for flexible configuration that is tailor-made for your operative needs.

With the WittyBox™ Series, Clue has met its objective of creating a solid platform to enable a new generation of intelligent systems capable of running AI technologies aboard small, medium and large aircrafts.

Calibre is the compact form factor within the WittyBox™ family. Thanks to its very reduced and robust enclosure, the Calibre series is born to feel at home in the hardest possible environments.

The system is highly configurable, thanks to the many options granted by multipurpose VPX modules engineered to successfully meet DO-178 and DO-254 certifications interconnected by a 5-slot VPX backplane with multi-gigabit point to point channels.

WittyBox™ Calibre is capable of over delivering as data-link processor, data/video recorder, data concentrator unit, communications server or mission computer.

Rock solid.
Lightweight.
High customization.
High performance.

Thanks to its very reduced and robust enclosure, the Calibre series is born to feel at home in the hardest possible environments.

High customization.
High performance.

Granted by multipurpose VPX modules engineered to successfully meet DO-178C and DO-254 certifications installed on a 5-slot backplane with multi-gigabit point to point channels.

TRUE

MULTI-CORE
PROCESSING

WittyBox™ Calibre equips the NXP QorlQ T2080 Microprocessor.

The NXP QorIQ® T2080 is a low-power, extended temperature range, 64-bit PowerPC Quad-core architecture running at 1.8 GHz targeted to RTCA DO- 254 applications. It provides the system with UARTs, Ethernet (with IEEE 1588 support), and PCIe interfaces. Additionally, NXP implement their trust technology to provide an array of security features that ensure confidentiality and integrity.

To support its high processing multi-core capabilities, Calibre can install memory up to 32 GB DDLR3 RAM with Error Correcting Code (ECC) to protect against memory errors.

Clue has added to its WittyBox™ Core boards a dedicated hardware accelerator to run embedded Deep Neural Network applications. Interfacing directly with other key components via the intelligent memory fabric, the Neural Engine is able to deliver bleeding-edge performance per Watt without facing common data flow bottlenecks encountered by other architectures.

This upgrade also provides imaging and vision engines including additional programmable SHAVE cores, updated and expanded vision accelerators, and a new native 4K ISP pipeline with support for up to 8 HD sensors connecting directly to the VPU.

This Intel-based core Neural Engine is capable of 150 GFLOPS (Giga Floating-Point Operations Per Second) and 4 TOPS (Trillion Operations per Second) at the expense of only 1W of power consumption.


Avionics Artificial Intelligence

150 TGFLOPS 150 TGFLOPS 150 TGFLOPS 150 TGFLOPS 150 TGFLOPS 150 TGFLOPS 150 TGFLOPS 150 TGFLOPS

4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS 4 TOPS

NATIVE 4K ISP PIPELINE NATIVE 4K ISP PIPELINE NATIVE 4K ISP PIPELINE NATIVE 4K ISP PIPELINE NATIVE 4K ISP PIPELINE NATIVE 4K ISP PIPELINE 4K ISP PIPELINE

Top-notch
storage

  • 1 TB
  • 2 TB
  • 3 TB
  • 6 TB

GPGPU
engine

  • 64 bit
    PowerPC
    Architecture
  • 1,8 GHz
    Quadcore
    Dual Thread
  • 42,3 Billion
    instructions
    per second

Besides other forms of hardware acceleration for heavy computing tasks, Clue is relying on GPGPU computing to boost the performance of WittyBoxTM Systems and meet the high processing requirements that characterize intelligent systems.

Clue’s WittyBox™ NVIDIA® Tesla® P6 based GPUs provide the best performance/cost ratio for applications requiring high computing muscle. Thanks to Tesla’s P6 small MXM form factor running at less than 90W power footprint, Clue’s new conduction cooled VPX GPU accelerates density-optimized, scale-out WittyBox™ servers to open new doors in avionic AI applications.

With its 2048 Cores reaching up to 6.2 TFLOPS, the NVIDIA® Tesla® P6 is designed to excel at data center like workloads, including deep learning, HPC, and graphics virtualization. It delivers higher graphics perfor- mance, improved energy efficiency and improved frame buffering.

The NVIDIA Tesla P6 is powered by the revolutionary NVIDIA Pascal™ architecture and is purpose-built to boost efficiency for scale-out servers running deep learning workloads, enabling smart responsive AI-based services. This rugged Pascal-based board runs in NVIDIA’s HPC mode which exposes a very large 16GB memory base address register (BAR) for direct access to the GPU memory from the CPU and other PCIe devices, and it includes Error Correcting Code (ECC) to protect against memory errors.

By reaching an energy efficiency up to 60x better than CPUs for deep learning inference workloads and low latency, WittyBox™ can offer the best performance per watt, which is fundamental in embedded systems.

Additionally, Clue’s WittyBox™ offers an AMD Radeon™ based solution for Safety Critical applications. This 3U OpenVPX high-performance embedded GPU board embeds an AMD Radeon™ E8860 powered GPGPU processing engine, with 2D and 3D graphics generation capabilities.

This is a GPGPU processing engine, certifiable up to DAL C (other levels upon request), developed for the most stringent aerospace and defense applications that require certification to DO-178C/DO-254. This pre-certified board can be delivered with all documentation, certification evidences and supporting artifacts required to prove compliance for avionics certification.

Featuring 6 independent video outputs and 2 video inputs, the GPU board can overlay captured video stream with generated 2D and 3D graphics elements and connect to the CPU through the VPX PCIe bus. All within safety-critical environments.

Best in class
interfaces

The NXP QorIQ® T2080 is a low-power, extended temperature range, 64-bit PowerPC Quad-core architecture running at 1.8 GHz targeted to RTCA DO- 254 applications. It provides the system with UARTs, Ethernet (with IEEE 1588 support), and PCIe interfaces. Additionally, NXP implement their trust technology to provide an array of security features that ensure confidentiality and integrity.


Ethernet
4x10
GBPS

ARINC
2x664
AFDX

ARINC
2x825
CANBUS

RS-
10x232
/422/485

MIL-STD-
2xSTD
-1553B

ARINC
32x429
 

ARINC
16xGPIO
 

USB
2x2.0
 

Security

Prevent attacks to stakeholder data and unauthorized system usage

WittyBox™ uses a combination of hardware and firmware solutions to create trusted systems. Trusted systems do what their stakeholders (OEMs, Service Providers, and Users) expect them to do and help to prevent attacks such as unauthorized access to stakeholder private data and unauthorized system usage, including zombification.

Attacking systems often look for patterns within network data to identify protocols or users; leveraging this information WittyBox™ can implement appropriate security or QoS policies to identify attempted security breaches and attacks. For this purpose, we use our own Pattern Matching Engine (PME), a type of regex hardware acceleration engine which scans data for patterns based on Regular Expression rules.

Cryptography is very computationally intensive. Our products include crypto accelerators to support system latency and bandwidth requirements. WittyBox™ crypto coprocessors address the public key offload required for data center and network security appliances as they scale into thousands of transactions per second. A scalable solution offers between 8 and 32k 2048-bit RSA transactions per second with power consumption from 4 to 10 watts.

WittyBox™ crypto coprocessors are designed to offload over a PCIe bus and can be used as a companion to QorIQ® communications processors such as the NXP T2080.

  • IPsec
  • SRTP
  • SSL/TLS
  • LTE PDCP
  • IEEE 802.1AE MACSec
  • IEEE 802.16e WiMax MAC layer
  • 3GPP RLC encryption/decryption
Calibre Martillo About Careers News