Description
Join us for an insightful webinar as we delve into the world of accelerated AI inference with tsunAImi. Untether AI was established with a vision to revolutionize the field of machine learning computation. Traditional architectures often suffer from significant energy consumption due to data movement during AI workloads. Addressing this challenge, Untether AI offers cutting-edge, energy-efficient hardware solutions such as the runAI family of devices and the tsunAImi® family of PCIe cards. These devices and our imAIgine® software support a wide range of AI workloads and deliver exceptional performance.
The runAI family of devices operates at an impressive speed of up to 502 TeraOperations per second in its “sport” mode. Alternatively, it can be configured for optimal efficiency, providing 8 TOPs per watt in “eco” mode. The tsunAImi family of PCIe cards is capable of supporting up to 2 PetaOps of AI inference with an efficiency of 8 TOPs per watt.
This webinar will provide comprehensive coverage on two key areas: ML Framework and Custom Kernel Writing and Lowering. In the ML Framework segment, topics such as graph ingestion, calibration and quantization, adding custom layers or operations, and quantization-aware training will be explored. The Custom Kernel Writing and Lowering portion will focus on utilizing the imAIgine CLI for network compilation, creating custom kernels, leveraging Tile Flow for HPC and kernel testing, incorporating directives into the compiler, and integrating custom kernels into the compilation flow. Additionally, a live demonstration showcasing the capabilities of tsunAImi will be presented, along with guidance on accessing early adopter programs.
Agenda
- ML Framework:
- Graph ingestion
- Calibration and Quantization
- Adding a custom layer or operation
- Quantization Aware Training
- Custom Kernel Writing and Lowering:
- Utilizing the imAIgine CLI for network compilation
- Creating custom kernels
- Leveraging Tile Flow for HPC and kernel testing
- Incorporating directives into the compiler
- Integrating custom kernels into the compilation flow
- Demo: tsunAImi and Early Access Details:
- Live demonstration of tsunAImi capabilities
- Guidance on accessing early adopter programs
Don’t miss this opportunity to gain valuable insights into accelerating AI inference with tsunAImi. Register now to secure your spot in the webinar.
Contact
If you have any comments or questions regarding the contents of this workshop, please contact Dr. Yassine Hariri at Hariri@cmc.ca.
Organizers
- Gaurav Singh, Director of Technical Marketing at Untether AI
Prior to Untether AI, Gaurav spent ~5 years in the Autonomous Vehicles Industry, first as a Machine Learning Researcher, and then later in the Business Strategy group—all at Ford. - Yassine Hariri, Sr. Scientist AI/ML, CMC Microsystems
Over 15 years of experience in advanced computing systems from the cloud to the very edge, with a focus on artificial intelligence, computer vision, video, image and sensor fusion workloads acceleration, FPGA-based prototyping, software stack, and domain-specific hardware architectures. Currently leading projects related to the specification, development, implementation, deployment, and support of the next generation of advanced computing infrastructure mainly FPGAs, GPUs, and Custom Hardware for AI applications. Dr. Hariri earned his B.A.Sc. in Computer Engineering from Ecole Marocaine des Sciences de l’ingénieur, Casablanca, Morocco, in 1998, and the M.S. and Ph.D. degrees from Ecole de Technologie Supérieure (ETS), Montreal, QC, Canada, in 2002 and 2008, respectively, all in electrical engineering.