Description
MasterGAN1-ST
Overview
The MasterGAN1-ST is a high-performance, single-threaded GAN (Generative Adversarial Network) accelerator designed for a wide range of applications, including image and video processing, natural language processing, and more.
Specifications
- Clock Speed: 1.2 GHz
- Core Count: 1
- Thread Count: 1
- Die Size: 10mm x 10mm
- Process Node: 12nm
- Memory Interface: 32-bit, 2GB
- Power Consumption: 5W
Performance
- Floating Point Operations Per Second (FLOPS): 120 GFLOPS
- Integer Operations Per Second (IOPS): 240 GIOPS
- Bandwidth: 20 GB/s
Features
- GAN Acceleration: Accelerates GAN workloads, including image and video generation, style transfer, and more
- Low Power Consumption: Operates at a low power consumption of 5W, making it suitable for battery-powered devices
- Small Form Factor: Compact die size of 10mm x 10mm, allowing for easy integration into small form factor devices
Applications
- Image and Video Processing: Accelerates image and video generation, style transfer, and more
- Natural Language Processing: Accelerates language translation, text summarization, and more
- Computer Vision: Accelerates object detection, segmentation, and more
Software Support
- Operating System: Linux, Windows, Android
- Programming Model: CUDA, OpenCL, TensorFlow
Package and Pinout
- Package Type: QFN
- Pin Count: 128
- Pinout:
- Power: 10 pins
- Ground: 10 pins
- Signal: 108 pins
Ordering Information
- Part Number: MasterGAN1-ST
- MOQ: 1000 units
- Lead Time: 6 weeks