GPU-accelerated Python with CuPy and Numba's CUDA

แชร์
ฝัง
  • เผยแพร่เมื่อ 25 ม.ค. 2025

ความคิดเห็น • 12

  • @arpitraval8516
    @arpitraval8516 2 ปีที่แล้ว +1

    I have one macbook and one windows laptop both has intel integrated graphics, without Nvidia graphics. i can't install cupy [Exception: Your CUDA environment is invalid. Please check above error log.] any other option to execute python script on GPU?

    • @overseashub241
      @overseashub241 2 ปีที่แล้ว +1

      You cannot done on integrated graphics

  • @aminreguig5714
    @aminreguig5714 ปีที่แล้ว

    Hi, I tried edit this code to get max of an array using cuda but it does not work and I dont know how to fix it
    can you please help me ?

  • @SkyRiderJavelin
    @SkyRiderJavelin 10 หลายเดือนก่อน +1

    well explained concepts and code

  • @prietjepruck
    @prietjepruck ปีที่แล้ว

    Wow, a really interesting video. I whish the code was givven so that I don't have to copy it from the screen. Thanks a lot for this tutorial.

  • @adfgadfgadfgadfgadf
    @adfgadfgadfgadfgadf 2 ปีที่แล้ว +1

    very helpful..thanks

    • @InfoWorld
      @InfoWorld  2 ปีที่แล้ว

      Glad it was helpful!

  • @atikzimmerman
    @atikzimmerman 3 หลายเดือนก่อน

    So, GPU is 5000x faster right? That's an estimate we've got

  • @overseashub241
    @overseashub241 2 ปีที่แล้ว +2

    😂😂😂😂😂😂 this code is run for 10**8 is 2.second,10**9 is 21 second, the data transferring time really takes only 0.1 s for 10**8 and 10**9 as under 1s .the real work is takes long the computation time is just happens suddednly because it is cpu happening not on gpu after the sieve done on gpu takes 1.9 s then only cpu capable of transferring data from gpu

  • @johnpsmith8132
    @johnpsmith8132 ปีที่แล้ว

    God damn it, I can't fucking install it in PyCharm.

  • @Steven-v6l
    @Steven-v6l 7 หลายเดือนก่อน

    Not all problems admit to parallel processing ... some problems are inherently sequential.
    here's an inherently parallel problem: if a pixel is inside a polygon, turn the pixel black.
    if you have 1,000,000 pixels and 1,000,000 processors, you can solve this problem really quickly.
    here's a problem that is not a parallel problem: compute f(27.194) where f(x) = cos(x^3 + 2x^2 + 13)
    But even if a problem is admits to parallel processing on GPUs; GPUs don't run Python.
    But you can translate a C program, into steps that WILL run on a GPU.
    Sometimes Python is just the *WRONG* language to use, to solve a problem.
    .