THANK YOU.

We have received your request for registration. We will send you a confirmation e-mail once your request has been approved.

SUBSCRIBE TO OUR NEWSLETTER.

Please register below to sign up for our CoreAVI Newsletter.

PLEASE SIGN IN TO ACCESS TO OUR RESOURCES.

Login below using the username and password created when you registered for CoreAVI resources.

Lost your password?

Not registered yet?

ComputeCore™

GPU Compute is the application of a modern GPU’s data-level parallelism through many-core SIMD engines to solve high volume, efficient computations. It is the basis for modern data processing, machine learning and computer vision systems. Computer vision consists of autonomous or augmented applications requiring a safe and robust computer vision framework.

CoreAVI’s ComputeCore™ is a suite of libraries that implement standard industry computational operations facilitating core building blocks for Safe AI, machine learning and autonomous systems using the Vulkan SC API. ComputeCore is offered in conjunction with CoreAVI’s VkCore® SC Khronos conformant Vulkan SC graphics and compute foundation layer, allowing applications to benefit from the performance gains and scalable capabilities offered by Vulkan. ComputeCore implements the industry standard Basic Linear Algebra Subprograms (BLAS) functions, a safety critical implementation of Fast Fourier Transforms (FFT), and a Neural Network (NN) library based on the Khronos NNEF (Neural Network Exchange Format) specification.

ComputeCore offers the following features and benefits:

  • SafeAI tools and framework that provides the building blocks to enable accelerated compute and visualization for autonomous systems using the Vulkan® SC API.
  • Provides safety critical deterministic inferencing of neural network models for a broad array of compute and AI oriented deployments.
    • Compatible with the interoperable ONNX Network Exchange Format enabling portability between popular Neural Network open-source model training frameworks and the Khronos approved ecosystem
    • Broad array of safety critical accelerated Vulkan SC based inference engines enabling the deployment of desktop-based expression models
  • Includes support for linear algebra and signal manipulation algorithms through the exposed BLAS and FFT compute interfaces.
    • Fundamental building blocks for computer vision and neural network deployments
    • Matrix manipulation in addition to image and signal enhancement capabilities
  • Supporting a broad range of Machine Learning, Autonomous, Compute and Visualization use cases servicing Automotive ADAS & L4/L5 Autonomy, Medical, Aerospace & Defense, and Industrial Robotics application domains.
  • Decoupled from hardware enabling cross platform portability and averting obsolescence overhead with the incumbent applications.
  • Enables applications to interact with both native Vulkan SC graphics and compute pipelines resulting in a high-performance parallel processing capability with low latency, low complexity, and easy to maintain visualization and compute solution.
  • Commercial Vulkan® build and execution option, facilitating off target (desktop) rapid / low-cost development.
  • Provides a safety critical certifiable alternative and facilitates the transition to Vulkan SC based Compute from OpenCL™ /CUDA®.
  • Designed from the ground up for real time and safety certification. Contains no open-source components and no 3rd party software.
  • Available with CertCore™26262/61508 (ISO 26262 Accredited Safety Assessment Certificate) and CertCore™178 (DO-178C / ED12-C Avionics) ASIL D / DAL A safety certification package.

Don’t see what you need? Please contact us at Sales@CoreAVI.com to determine if we can help support your certifiable application needs.