GPU Programming NSM Nodal Centre for Training in HPC and AI

NSM Nodal Centre for Training in HPC and AI is organizing an online course on GPU Programming. The course is free, for everyone. The interested may register for the course here. The course is taught via recorded lectures and doubt sessions. The course is derived from a similar course taught at IIT Madras in parallel. The topics are listed below.

Introduction Computation Memory Synchronization Functions Support Streams Topics and Case study Who may attend this course? if you have interest in learning to program GPUs students from science and engineering company personnel interested in programming GPUs scientists and domain experts Who should NOT attend this course? if you never heard of a graphics processing unit if you do not know programming

FEEDBACK FOR GPU PROGRAMMING I shall undergo Thread Divergence As I launch my Feedback Kernel in a poetic way, Thank you Sir, for being a Host par excellence To me, a Thread from another Device, I say. The Stream of your lectures was appealing, Each day I was hooked, in Pinned Memory, Awaiting your videos on the PCI Express bus each morning, All your programs I did diligently cudaMemcpy. Owing to Coalescing, I couldn't just watch one lecture, But had to make Strided Access to subsequent ones too; Till I watched them all -- one big Vector! And so in Global DRAM, I want to thank you! You patiently resolved all Race Condition Of doubts and questions without making Lost Update, You encouraged interaction and Synchronization, In everyone's Shared Memory, you earned a place great! As a Warp Representative from this class, I perform an Inclusive Scan of all you taught, You did Reduction of concepts like no one has; atomicAdd(&likes, 1) to all your analogies' lot. The Prefix Sum of my feedback is this: You taught in a SIMD fashion, With a Global Barrier to ensure no one did miss, Thus, __all(Prof Rupesh is awesome) returns 1. -- Ullas Aparanji, IISc

Rupesh Nasre. Faculty, CSE, IIT Madras Coordinator, NSM Nodal Centre