Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

ggml : add ggml_soft_max_ext #4256

Merged
merged 14 commits into from
Dec 1, 2023
Merged

ggml : add ggml_soft_max_ext #4256

merged 14 commits into from
Dec 1, 2023

Commits on Nov 29, 2023

  1. metal : implement soft_max_ext

    @ggerganov
    ggerganov committed Nov 29, 2023
    Configuration menu
    Copy the full SHA
    e89597c View commit details
    Browse the repository at this point in the history

  2. cuda : implement soft_max_ext

    @ggerganov
    ggerganov committed Nov 29, 2023
    Configuration menu
    Copy the full SHA
    88519fb View commit details
    Browse the repository at this point in the history

  3. ggml : implement soft_max_ext (CPU)

    @ggerganov
    ggerganov committed Nov 29, 2023
    Configuration menu
    Copy the full SHA
    6a66f69 View commit details
    Browse the repository at this point in the history

  4. batched-bench : print threads 

    ggml-ci
    @ggerganov
    ggerganov committed Nov 29, 2023
    Configuration menu
    Copy the full SHA
    390a445 View commit details
    Browse the repository at this point in the history

  5. metal : simplify soft_max encoding 

    ggml-ci
    @ggerganov
    ggerganov committed Nov 29, 2023
    Configuration menu
    Copy the full SHA
    580fe20 View commit details
    Browse the repository at this point in the history

Commits on Nov 30, 2023

  1. cuda : use 512 threads for soft_max instead of 32

    @ggerganov
    ggerganov committed Nov 30, 2023
    Configuration menu
    Copy the full SHA
    ebd062b View commit details
    Browse the repository at this point in the history

  2. ggml : update soft max cpu

    @ggerganov
    ggerganov committed Nov 30, 2023
    Configuration menu
    Copy the full SHA
    c7c8dab View commit details
    Browse the repository at this point in the history

  3. cuda : do warp-based block reduce

    @ggerganov
    ggerganov committed Nov 30, 2023
    Configuration menu
    Copy the full SHA
    62532c0 View commit details
    Browse the repository at this point in the history

  4. cuda : increase max block size to 1024

    @ggerganov
    ggerganov committed Nov 30, 2023
    Configuration menu
    Copy the full SHA
    6b86bcf View commit details
    Browse the repository at this point in the history

  5. cuda : fix warp reduction initialization of shared mem

    @ggerganov
    ggerganov committed Nov 30, 2023
    Configuration menu
    Copy the full SHA
    68e02c0 View commit details
    Browse the repository at this point in the history

  6. metal : warp-based reduction for soft max kernel

    @ggerganov
    ggerganov committed Nov 30, 2023
    Configuration menu
    Copy the full SHA
    55717c9 View commit details
    Browse the repository at this point in the history

  7. metal : warp-based reduce for rms_norm

    @ggerganov
    ggerganov committed Nov 30, 2023
    Configuration menu
    Copy the full SHA
    c4db592 View commit details
    Browse the repository at this point in the history

Commits on Dec 1, 2023

  1. metal : simplify soft max kernel 

    ggml-ci
    @ggerganov
    ggerganov committed Dec 1, 2023
    Configuration menu
    Copy the full SHA
    d9c8fa3 View commit details
    Browse the repository at this point in the history

  2. alloc : fix build with debug

    @ggerganov
    ggerganov committed Dec 1, 2023
    Configuration menu
    Copy the full SHA
    eb594c0 View commit details
    Browse the repository at this point in the history