Fundamentals of GPU Architecture: Programming Model Part 1

Показать описание

In this video we look at the basics of the GPU programming model!

Рекомендации по теме

Комментарии

23:40 I think it should be "block/thread index" instead of "block/thread ID", hence the extra x.

jamesmurdza

Thank you for this amazing walk-through on Chapter 2 of Synthesis Lecture. I had a question:
The code mentioned in Figure 2.2, line 18: Why are we copying array x instead of array y? I am asking because the final computation is stored in array y.

rishabhjain

I come from a computer graphics background (3d rendering engine OpenGL) and I am learning GPU architecture to prep for my interviews. Will General Purpose GPU architecture in the tutorials knowledge will suffice or there is a different genre of GPU architecture specifically for Graphics. Sorry for my dumb question but ya I am a newbie..

podilasahithi

@~23:00, what could happen if, by mistake or intent, one would pass, to the kernel, the host pointers instead of the device pointers ?

biocode

Hi nice course ! I had a question :
why do we use multi-threaded blocks? We could use blocks composed of only one thread, it would avoid us to have threads that are useless, right?
Moreover how is it done in the example shown if, we have less threads than we need to iterate, for example the size of the matrix is 10 and we have only 1 block of 2 threads, the program would not work in this case no? Maybe we would have had to use a for loop with an interval ?

mouldor-

Is it possible to have a programmable ROM in the GPUs for the starter code instead of using CPU to start the GPU?

apurvechawda

Hey, I didn't understand CudaMalloc. The pointer d_x and d_y were created on the Host's memory. Then they are being used to allocated Memory on the Device and these will have the location of that memory. So pointer stored in Host's memory have address to allocated space in device's memory and these can only be accessed by the GPU for computation and storing the result in the allocated space in it's memory which then will be copied to the Host. Is this correct?

abhishektyagi

Fundamentals of GPU Architecture: Programming Model Part 1

Fundamentals of GPU Architecture: Programming Model Part 1

Nvidia CUDA in 100 Seconds

CPU vs GPU | Simply Explained

GPUs: Explained

Fundamentals of GPU Architecture: Introduction

How do Graphics Cards Work? Exploring GPU Architecture

Intro to CUDA (part 1): High Level Concepts

Intro to CUDA - An introduction, how-to, to NVIDIA's GPU parallel programming architecture

CPU vs GPU vs TPU vs DPU vs QPU

Introduction | GPU Programming | Episode 0

CUDA Part A: GPU Architecture Overview and CUDA Basics; Peter Messmer (NVIDIA)

Modern GPU Architecture | GPU Programming

Fundamentals of GPU Architecture: SIMT Core Part 1

An Introduction to GPU Architecture and Programming Models | Tim Warburton, Virginia Tech

Graphics Processing Unit (GPU)

CUDA Programming Course – High-Performance Computing with GPUs

Introduction to GPU Architecture and Programming Models ǀ Tim Warburton, Virginia Tech

Fundamentals of GPU Architecture: Programming Model Part 2

How Does A GPU Work? | PC Guide EXPLAINED

Computer Architecture - Lecture 25: GPU Programming (ETH Zürich, Fall 2020)

An Intro to GPU Architecture and Programming Models I Tim Warburton, Virginia Tech

Introduction to the GPU architecture, Karakasis

GPUs Explained for Beginners | What is a GPU?

GPU Architectures and New Programming Model Features | Nikolai Sakharnykh, NVIDIA