Developments in AI Inferencing and Implications for Edge and Edge Infrastructure Processors
This IDC Market Perspective discusses different AI network types, implications for processors, key types of processors for AI, and methods for implementation. AI developments continue to change the semiconductor market. New methods for inferencing, changing bit precision for inference, new semiconductor manufacturing techniques for heterogeneous SoCs, and other developments must be taken into consideration in semiconductor design and choice."AI continues to evolve rapidly, and companies need to evaluate the development of semiconductor processors, taking new networks and the compute required into consideration," says Nina Turner, research director, IDC Enabling Technologies and Semiconductor team.
Please Note: Extended description available upon request.
- Executive Snapshot
- New Market Developments and Dynamics
- Types of AI Networks and Mechanisms
- Feedforward Neural Networks
- Recurrent Neural Networks
- Convolutional Neural Networks
- Generative Adversarial Networks
- Attention Mechanism
- Transformer Neural Networks
- Graph Neural Networks
- Other Compute Requirements
- Parallel Processing
- Memory Access
- Interconnect (I/O)
- Precision
- Architectures Addressing the Challenges
- Digital Signal Processor
- Graphics Processing Unit
- Neural Processing Unit
- Field-Programmable Gate Array
- Methods of Adding in AI Blocks
- Integration
- Discrete Accelerator
- Multi-Die SoCs
- Advice for the Technology supplier
- Learn More
- Related Research
- Synopsis
Research Assistance
Download our eBook: How to Succeed Using Market Research
Learn how to effectively navigate the market research process to help guide your organization on the journey to success.
Download eBook