+1 -1 +16
Vote on this proposal

GPU Programming terminology demystified before you jump into CUDA programming

by Narahari Allamraju (speaking)

Section
Software Development Tools
Technical level
Beginner

Objective

CUDA and GPU programming are the buzz words - but how many of us understand how exactly this works? We all learnt CPU architecture in college and understand how the various registers etc work - but not all of us know about the inner workings of GPU.

This talk will aim to help everyone understand the basic terminologies and concepts of GPU programming.

There is no basic skill level required for this talk.

Description

How much do we know about how a GPU works? We know that a GPU can massively parallelize tasks for us - but how does it do it? How is the architecture of a GPU different from that of a CPU? And how does this work in our favor?

Lets dig in and understand all this, right from the basics of what these GPU's were originally built to achieve and how that is changing our programming lives these days.

Requirements

No basic requirements for this talk - anyone can easily understand.

Speaker bio

Hari is a Java developer with nearly 9 years of experience building large scale messaging platforms and electronic trading applications at an investment bank. He has been interested in Python for the last couple of years and has been using that in his daily assignments to prototype Java applications on Jython, develop basic Django web applications and to automate some routine tasks.

With my background in electronics engineering, I will be able to explain the GPU architecture in general terms and more clearly to even new comers.

Comments


  • 1

    [-] Anand B Pillai 565 days ago

    Wondering what this talk has to do with Python. Are you planning to talk on something like pyCUDA or similar libraries ?


  • 1

    [-] Narahari Allamraju 565 days ago

    Hi Anand - I was attending a session on pyCUDA at PyconSG and realized that there are many differences between GPU and CPU architectures (corees vs kernels) etc which need to be very clear before we attempt doing something with pyCUDA - so I put in this proposal to get beginners up to speed on these terminologies. I see that there are talks on pyCUDA - so for beginners this will help prepare them for better understanding of those talks.

Login with Twitter or Google to leave a comment →