Intel® Time Coordinated Computing and AI at the Edge
Overview
A major transformation towards software-defined solutions and AI is happening at the edge. It's driven by the need for flexibility and scalability, changes in the workforce, automation, advances in edge-based AI, and more. A key element of this transformation is the capability to support real-time workloads along with best-effort workloads (mixed-criticality) on the same system. Moreover, emerging software-defined systems and AI applications are distributed, involving multiple computing nodes (for example, multiple sensors, controllers, and actuators) that perform coordinated actions based on a shared understanding of time. This distributed computing paradigm, Time Coordinated Computing, is crucial across edge applications, including real-time systems, industrial automation, distributed displays, large-scale simulations, and digital twins using edge AI.
Edge platforms from Intel include optimized hardware capabilities and a strong software ecosystem to support mixed-criticality workloads and ensure real-time performance, delivering a notable Time Coordinated Computing experience.
Intel Time Coordinated Computing Experience at the Edge
Edge platforms from Intel deliver a notable Intel® Time Coordinated Computing (Intel® TCC) experience with high-performance compute and real-time guarantees by:
- Prioritizing real-time workloads access to cache, memory, and networking resources
- Minimizing disruption from other workloads
- Optimizing performance for both real-time and non-real-time workloads
- Supporting availability in both native and virtualized environments
- Providing precise and synchronized timing within the platform and across wired and wireless networks
- Offering support by a large software ecosystem and standards-based interoperable connectivity
What Is Intel TCC?
Intel® TCC refers to the comprehensive set of optimizations throughout the platform and a network stack from Intel silicon to network interface cards that enable a notable TCC experience for our customers that includes:
- Silicon and platform software optimizations.
- End-point connectivity to support IEEE time-sensitive networking (TSN) standards that reduces latencies and improves clock synchronization for real-time workloads distributed across wired and wireless networks.
Prioritization of Real-Time Workloads
When specific real-time workloads must be completed within a set deadline to avoid critical system failure, the ability to prioritize some workloads over others is paramount.
High Predictability and Reliability
Real-time workloads process data in defined, predictable time frames, enhancing the reliability of safety and time-critical systems.
More Precise Timing
Real-time systems and time-aware applications are designed to perform tasks at precise times, within strict deadlines (down to microseconds).
Deterministic Low Latency
Applications and distributed systems can coordinate actions across wired and wireless networks with high precision and determinism.
FAQs
Frequently Asked Questions
This is a compute workload that needs to be run within predictable and specific time constraints. Such workloads may be part of many applications spanning from process automation to production assembly lines, robotics, transportation systems, healthcare, and live performances. A few examples of real-time workloads include:
- A real-time controller driving a manufacturing process.
- A control system within a vehicle or aircraft.
- A real-time audio/video mixing application that's part of a live performance.
Real-time workloads may also include AI components such as real-time data fusion of multiple sensors in an autonomous robot or vehicle used as input to safety and time-critical decisions.
A mixed-criticality environment refers to a computing and networking environment that can run multiple workloads with different performance requirements. Some workloads may have time-critical performance constraints, and other workloads may only need best-effort performance. As more and more systems and applications transition from dedicated hardware to software-defined systems, the need to support mixed-criticality workloads on the same general-purpose compute platform and network is becoming a fundamental requirement across multiple markets.
Intel TCC includes several optimizations to ensure real-time performance in a mixed-criticality computing and networking environment. Some of the capabilities and optimizations include: precision time coordination, power-state transition optimizations, memory/cache allocation, interrupt request (IRQ) optimizations, virtual channels, Intel® Speed Shift Technology for edge compute (enables assignment of processor performance as needed), and support for TSN.
Key Use Cases
Edge AI
Flight Control
Industrial Controls Applications
Video Wall
Real-Time Audio & Video Processing
Medical Imaging
Robotics
Development and Testing Resources
Intel works with the broader community to enable a strong software ecosystem to take advantage of platform optimizations and support the Intel Time Coordinated Computing experience. Developers and users who build applications that need real-time performance can use the following documentation:
- TCC Tutorial: A guide on optimizing Linux* real-time performance on Intel CPUs is available on GitHub*.
- Canonical* Guide to Real-Time Linux: Recommended reading as an introduction to Linux real-time capabilities.
- Real-Time Ubuntu* Documentation: Optimizing real-time performance on Intel CPUs is a collaborative tutorial between Intel and Canonical.
- Linux Real-Time Communication Testbench: An open source tool for validating and testing real-time applications and TSN capabilities on Linux systems is available on GitHub.
- TSN Reference Applications: A set of TSN samples and reference applications from Intel is available on GitHub.
Featured Platforms
For a list of edge computing offerings at Intel, see Embedded Processors.
Customer and Partner Success Stories
Real-Time Solutions Hypervisor Video
"The introduction of the RTS Hypervisor on the latest Intel platforms facilitates new possibilities for real-time applications ranging from robotics and smart cars to factory automation and utility load management."
— Michael Reichlin, general manager of Real-Time Systems
Wind River VxWorks* Operating System: Real-Time and Functional Safety Solutions on Intel-Based Industrial Platforms
Documentation
Get design documents, drivers, datasheets, release notes, and more for IoT real-time solutions.