[HN Gopher] Ask HN: Best way to learn GPU programming?
       ___________________________________________________________________
        
       Ask HN: Best way to learn GPU programming?
        
       I'd like to learn GPU programming but I'm having difficulty finding
       high-quality resources. I tried a class at coursera and was
       severely disappointed by both quality and content.  What are the
       best resources for learning things like GPU architecture, CUDA,
       Triton, etc?  My goal is to do be able to do something like take a
       description of Flash Attention and implement it from scratch, or
       optimize existing CUDA code.
        
       Author : hazard
       Score  : 19 points
       Date   : 2024-01-01 22:13 UTC (2 days ago)
        
       | Baldbvrhunter wrote:
       | It would also mean learning Julia, but you can write GPU kernels
       | in Julia and then compile for NVidia CUDA, AMD ROCm or IBM
       | oneAPI.
       | 
       | https://juliagpu.org/
       | 
       | I've written CUDA kernels and I knew nothing about it going in.
        
         | madsciencetist wrote:
         | While I am a fan of Julia and its GPU module, using such an
         | easy environment will really limit what you are able to learn.
         | NVIDIA provides some great optimization tools (NSight Systems
         | and NSight Compute) which help you optimize your kernel
         | execution (fuse kernels, hide latency, use execution graphs)
         | and CUDA code (take advantage of memory layout, use warp
         | intrinsics, maximize throughput). These tools map to C++/CUDA
         | source code and let you rapidly address bottlenecks - most of
         | which may be on the host side.
        
       | Kon-Peki wrote:
       | For CUDA specifically, there is a fairly large set of sample code
       | that used to be installed when you installed CUDA. But now I
       | think it's on the Nvidia GitHub page; you've got to download it
       | yourself.
       | 
       | The Nvidia dev blog has some easy to follow tutorials, but they
       | don't get very complicated.
       | 
       | Nvidia also has a learning platform which offers fairly decent
       | courses at a cost. You get a certificate for finishing.
       | 
       | You'll find some books out there with good reputations.
       | Ultimately, this is an area that leans heavily toward paying
       | money for good quality learning materials.
        
         | hazard wrote:
         | The best course by NVidia looks like "Fundamentals of
         | Accelerated Computing with CUDA C/C++" which I think used to be
         | publicly available, but is now offered "By invitation only"
        
           | Kon-Peki wrote:
           | It appears as though that course has been replaced with
           | "Getting Started with Accelerated Computing in CUDA C/C++"
           | which is $90 as a self-paced online class with a certificate
           | at the end.
           | 
           | The course you are referring to is instructor-led and costs
           | $10,000 to have them set up, for up to 20 people.
        
       ___________________________________________________________________
       (page generated 2024-01-03 23:02 UTC)