Boosting Semiconductor Quality Control with iEPF-9030S and PEGAAi

In IC substrate manufacturing, even a small number of false positives from Automated Visual Inspection (AVI) can lead to unnecessary re-inspection, slowing output and tying up valuable manpower. To address this, a leading semiconductor packaging manufacturer implemented an AI-driven secondary inspection system powered by the ASRock Industrial iEPF-9030S Series and PEGATRON’s PEGAAi platform. The system automatically filters out non-defective units in real time, reducing manual re-inspection workload by up to 70% and freeing operators for higher-value tasks. Precision defect classification has improved inspection accuracy, stabilized yield, and ensured consistent quality across production lines. Scalable by design, PegaAi has been rolled out to more than 400 stations in 13 factories worldwide, creating a unified, AI-enabled quality control process that enhances productivity and accelerates smart manufacturing at scale.

 
 

Scalable Edge AI Systems

iEPF-9030S Series Expandable Edge AI System for Light AI Workloads

    - High-Performance Compute, powered by Intel® 14th Gen Core™ processors with up to 96GB DDR5 memory.
    - Compatible with a single NVIDIA RTX™ GPU for AI workloads and light to moderate compute tasks.
    - Operates from -40°C to 75°C with robust power and rugged design.
    - Ideal for visual inspection, robotics control, and smart factory automation.

iEPF-10000S Series Expandable Edge AI System for Medium AI Workloads

    - Powered by Intel® Core™ Ultra processors (Arrow Lake-S) for high performance edge AI
    - Accommodates up to two NVIDIA® GPUs for medium intensity AI workloads.
    - Supports up to 256GB DDR5 for large AI models and data intensive applications.
    - Ideal for real-time analytics, multilingual AI chatbots, and computer vision in public infrastructure.

iEPF-11000S Series Expandable Edge AI System for Heavy AI Workloads

    - Equipped with Intel® Xeon® W-3500/2500 Series processors for mission-critical AI workloads.
    - Handles up to four NVIDIA® GPUs for high-intensity parallel processing.
    - Supports up to 2TB DDR5 ECC RDIMM for large AI training datasets.
    - Suited for large scale AI training, complex visual inspection, and autonomous system control.