Xilinx is now part ofAMDUpdated Privacy Policy

Zebra by Mipsology ® Best performing AI software for computing neural network inference

Zebra combines high performance with unprecedented ease of use and is AMD Xilinx’s preferred AI accelerator for computing neural networks for image-based AI applications.

Zebra seamlessly replaces GPU/CPU to compute any image based neural network faster and with lower power consumption. And with Zebra, there is no need to retrain or make any changes to your network or application. Zebra deploys with a simple Linnux command, so you don’t need FPGA knowledge to use Zebra.

Vendor: Mipsology

Try or Buy

Begin a free trial and run the application on VMAccel Cloud.


Purchase solution


Mipsology Zebra Neural Network Computing Software Demo 

These interactive demos allow users to see live results of Zebra computing neural network inference based on Xilinx hardware. Users can select the neural network and the Xilinx hardware and see live results. Each result can be inspected directly in a web browser for each image processed.

Get Started: Zebra Demo

Step 1 Visit Zebra performance demo
Step 2: Select your neural network
Step 3: Select your Xilinx hardware
Step 4: See live results of Zebra inference computation using your selections. 
Step 5: Fill out the Zebra inquiry form to get information on how to download Zebra

Get Started: Live Zebra on VCK5000

Step 1: Try Zebra VCK5000 Live demo
Step 2: Drag and drop your photos in the white text box on the top of the screen
Step 3: See live results of Zebra inference computation on the VCK5000 using your selections.
Step 4:  Fill out the Zebra inquiry form to get information on how to download Zebra


Getting Started

Follow the instructions based on your deployment method.

VMAccel Cloud

1.

Overview

This is an Alpha release with following goals:

Demonstrates Zebra functionality on VCK5000 board.

  • Demonstrates Zebra software Ease-of-Use   
  • Accelerate post-training Convolution Neural Network (CNN) model without any structural modification. NO pruning or re-training of the model.
  • Automatic and in-line quantization/calibration. No offline or separate compilation tool.
  • Performance for many CNNs is high; but not best possible. Details in later section.
  • Accuracy for most models is high; but not best possible.

2.

License

VMAccel Zebra virtual machines  (VM) are  pre-configured  with  software  license.  This license is not designed for production deployment. Any CNN inference running continuously for more than 15 minutes will experience significant slowdown in execution.


3.

Contact & Support

Please email Support for questions, concerns, technical help or discuss your project’s unique requirements. Please email License Team for questions related to Zebra License.


4.

Requirements

VMAccel Zebra instances are pre-configured with all required software and hardware. Only requirement on client side is a computer with internet connection and web browser. 


5.

VMAccel Cloud Access

To gain access to VMAccel cloud, please fill the form. For any questions or concerns, please contact our support.


6.

Launching Zebra VM Instances

Once you have the access credentials  from  VMAccel,  please follow “Getting Started” section.

Summary:

  • In a web browser navigate to:  https://xilinx2.vmaccel.com/dashboard/project/
  • On left hand side: Click on “Instances”
  • On right hand side: Click on “Launch Instance”
  • Follow GUI instructions/options for configuration
    IMP: Use following details when creating Zebra VM instance
    • Source: Secure Boot Image=“Mipsology Zebra VCK5000 ES1”
  • Flavor=“Mipsology Zebra VCK5000-ES1.1 (16.32.128)”
  • Network = “mipsology_local”

For advanced features like enabling external sshaccess, follow instructions.

For more information, please refer to Getting Started Guide.


7.

Results

Use this demo for two ways to see compute power of Zebra on the Zebra certified Xilinx hardware of your choice. You can either select your neural network and hardware and watch Zebra compute OR you can drag and drop your own local images (or images from a web search) and watch Zebra compute your selected images.