Skip to content

This Server.py and Client.py can provide send job to multiple GPU Server to run SFEGO in GPU which can make better throughput.

Notifications You must be signed in to change notification settings

CardLin/SFEGO_PyOpenCL_Service

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Spatial Frequency Extraction using Gradient-liked Operator (SFEGO) Service

PyOpenCL Version

Introduction

Hardware Requirement

  • Require GPU on Server to execute OpenCL Kernel Code

  • Recommend to use NVIDIA GPU with 1GB+ VRAM (VRAM usage is depend on Image Size)

  • AMD Integrated GPU and Intel Integrated GPU can also run this project

  • Although It can also run OpenCL on CPU mode but even the Intel Integrated GPU is faster than high-end CPU

Execution

  • Modify PLATFORMS = [(0,0,4),(0,1,8)] in Server.py which is (Platform_ID, Device_ID, Worker_Count)

  • I have two AMD GPU on this server. [(0,0,4),(0,1,8)] means run 4 thread on (Platform_ID=0, Device_ID=0) and 8 thread on (Platform_ID=0, Device_ID=1)

  • python Server.py

  • Modify ServerList = [ ("127.0.0.1", 8888, 12), ("192.168.1.33", 8888, 8) ] in Client.py which is ((IP, port, ExecuteCount))

  • ExecuteCount is concurrent thread that how many socket connect to specific Server, you can set this number as Worker_Count on server

  • Client.py support send image to different server to increase throughput

  • Modify IN_Folder for read image and OUT_Folder for save Spatial Frame

  • python Client.py

About

This Server.py and Client.py can provide send job to multiple GPU Server to run SFEGO in GPU which can make better throughput.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published