StarCore Appliance / Aerospace AI Large Model Integrated Appliance
    05

    Aerospace AI Large Model Integrated Appliance

    Aerospace AI Large Model Integrated Appliance

    KY-StarCore-AGENT | Safer Space · More Reliable Satellites
    Aerospace AI Large Model Integrated Appliance

    Product Overview

    Designed for the commercial aerospace sector, it deeply integrates large model training, on-orbit real-time inference, and multi-source aerospace data processing capabilities to efficiently accomplish large-scale constellation intelligent management, remote sensing data automatic interpretation, and space situational awareness tasks.

    Application Scenarios

    Serves satellite operators, data service providers, and research institutions, providing closed-loop support from intelligent computing and mission planning to data product generation.

    Core Advantages

    Highly integrated, with hardware-software co-optimization and an open architecture that significantly enhances the automation level and decision-making efficiency of commercial aerospace missions.

    Large Model Training Real-time Inference Constellation Management Remote Sensing Interpretation 200+ TOPS

    AI Capabilities

    Deeply integrating large model training with on-orbit real-time inference to empower commercial aerospace intelligence

    AI Capabilities

    Large Model Training

    Supports localized training and fine-tuning of aerospace large models, keeping data secure and controllable within the domain while meeting customized requirements

    On-orbit Real-time Inference

    Inference computing power up to 275 TOPS (INT8), meeting demanding on-orbit real-time intelligent decision-making needs with low latency and high throughput

    Multi-source Data Processing

    Integrates remote sensing, orbital, situational awareness, and other multi-source aerospace data for automatic interpretation and intelligent analysis with closed-loop support

    Open Architecture

    Standardized interfaces and protocols for flexible expansion, seamless integration with existing systems, and reduced migration costs

    Product Advantages

    275 TOPS Computing Power

    Optional up to 275 TOPS inference computing power, meeting the most demanding on-orbit intelligent decision-making requirements

    Data Never Leaves the Domain

    Localized training and inference keep data within the device at all times, ensuring data security and privacy

    Hardware-Software Synergy

    Deep optimization between algorithms and hardware to fully exploit computing performance, achieving low latency and high throughput

    Automated Decision Making

    Significantly improves the automation level and decision-making efficiency of commercial aerospace missions while reducing manual intervention

    Technical Specifications

    Model KY-StarCore-AGENT
    Manufacturer Beijing Kaiyun United Information Technology Group Co., Ltd.
    CPU 12×Cores Arm® Cortex®-A78AE v8.2 64-bit CPU, 3MB L2 + 6MB L3, CPU Max Frequency 2.2 GHz
    Inference Computing Power DL Accelerator + VS Accelerator: 2×NVDLA v2, PVA v2.0, up to 200 (standard) / 275 (optional) TOPS (INT8)
    GPU 2048×Cores NVIDIA Ampere Architecture GPU 1.3GHz with 64×Tensor Cores
    Dimensions (L×W×H) Main Chassis: 282 × 192 × 75 mm
    Operating Temperature Maximum range -25°C to 70°C (with external chassis fan)
    Storage Temperature -40°C~85°C
    Storage Humidity 10~90% @ 40°C, non-condensing
    Back to Product List