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

WIP: [CUDA] New CUDA version #4528

Closed
wants to merge 70 commits into from
Closed

WIP: [CUDA] New CUDA version #4528

wants to merge 70 commits into from

Commits on Apr 20, 2021

  1. new cuda framework

    @shiyu1994
    shiyu1994 committed Apr 20, 2021
    Configuration menu
    Copy the full SHA
    94aed50 View commit details
    Browse the repository at this point in the history

Commits on Apr 22, 2021

  1. add histogram construction kernel

    Yu Shi committed Apr 22, 2021
    Configuration menu
    Copy the full SHA
    18df6b2 View commit details
    Browse the repository at this point in the history

Commits on Apr 29, 2021

  1. before removing multi-gpu

    Yu Shi committed Apr 29, 2021
    Configuration menu
    Copy the full SHA
    9b21d2b View commit details
    Browse the repository at this point in the history

  2. new cuda framework

    Yu Shi committed Apr 29, 2021
    Configuration menu
    Copy the full SHA
    634a4f1 View commit details
    Browse the repository at this point in the history

Commits on May 6, 2021

  1. tree learner cuda kernels

    Yu Shi committed May 6, 2021
    Configuration menu
    Copy the full SHA
    23bcaa2 View commit details
    Browse the repository at this point in the history

Commits on May 7, 2021

  1. single tree framework ready

    Yu Shi committed May 7, 2021
    Configuration menu
    Copy the full SHA
    6c14cd9 View commit details
    Browse the repository at this point in the history

Commits on May 9, 2021

  1. single tree training framework

    Yu Shi committed May 9, 2021
    Configuration menu
    Copy the full SHA
    aa0b3de View commit details
    Browse the repository at this point in the history

  2. remove comments

    Yu Shi committed May 9, 2021
    Configuration menu
    Copy the full SHA
    bc85ced View commit details
    Browse the repository at this point in the history

Commits on May 10, 2021

  1. boosting with cuda

    Yu Shi committed May 10, 2021
    Configuration menu
    Copy the full SHA
    18d957a View commit details
    Browse the repository at this point in the history

Commits on May 13, 2021

  1. optimize for best split find

    Yu Shi committed May 13, 2021
    Configuration menu
    Copy the full SHA
    28186c0 View commit details
    Browse the repository at this point in the history

Commits on May 14, 2021

  1. data split

    Yu Shi committed May 14, 2021
    Configuration menu
    Copy the full SHA
    60c7e4e View commit details
    Browse the repository at this point in the history

Commits on May 17, 2021

  1. move boosting into cuda

    Yu Shi committed May 17, 2021
    Configuration menu
    Copy the full SHA
    57547fb View commit details
    Browse the repository at this point in the history

Commits on May 27, 2021

  1. parallel synchronize best split point

    Yu Shi committed May 27, 2021
    Configuration menu
    Copy the full SHA
    608fd70 View commit details
    Browse the repository at this point in the history

Commits on Jun 1, 2021

  1. merge split data kernels

    Yu Shi committed Jun 1, 2021
    Configuration menu
    Copy the full SHA
    277be8b View commit details
    Browse the repository at this point in the history

  2. before code refactor

    Yu Shi committed Jun 1, 2021
    Configuration menu
    Copy the full SHA
    ffcf765 View commit details
    Browse the repository at this point in the history

Commits on Jun 2, 2021

  1. use tasks instead of features as units for split finding

    Yu Shi committed Jun 2, 2021
    Configuration menu
    Copy the full SHA
    a58c1e1 View commit details
    Browse the repository at this point in the history

  2. refactor cuda best split finder

    Yu Shi committed Jun 2, 2021
    Configuration menu
    Copy the full SHA
    72d41c9 View commit details
    Browse the repository at this point in the history

  3. fix configuration error with small leaves in data split

    @shiyu1994
    shiyu1994 committed Jun 2, 2021
    Configuration menu
    Copy the full SHA
    f7a7658 View commit details
    Browse the repository at this point in the history

  4. skip histogram construction of too small leaf

    @shiyu1994
    shiyu1994 committed Jun 2, 2021
    Configuration menu
    Copy the full SHA
    b6efd10 View commit details
    Browse the repository at this point in the history

Commits on Jun 3, 2021

  1. skip split finding of invalid leaves 

    stop when no leaf to split
    @shiyu1994
    shiyu1994 committed Jun 3, 2021
    Configuration menu
    Copy the full SHA
    6f4e39d View commit details
    Browse the repository at this point in the history

Commits on Jun 4, 2021

  1. support row wise with CUDA

    @shiyu1994
    shiyu1994 committed Jun 4, 2021
    Configuration menu
    Copy the full SHA
    4072bb8 View commit details
    Browse the repository at this point in the history

  2. copy data for split by column

    @shiyu1994
    shiyu1994 committed Jun 4, 2021
    Configuration menu
    Copy the full SHA
    88ecde9 View commit details
    Browse the repository at this point in the history

Commits on Jun 8, 2021

  1. copy data from host to CPU by column for data partition

    @shiyu1994
    shiyu1994 committed Jun 8, 2021
    Configuration menu
    Copy the full SHA
    dec7501 View commit details
    Browse the repository at this point in the history

  2. add synchronize best splits for one leaf from multiple blocks

    @shiyu1994
    shiyu1994 committed Jun 8, 2021
    Configuration menu
    Copy the full SHA
    2dccb7f View commit details
    Browse the repository at this point in the history

Commits on Jun 9, 2021

  1. partition dense row data

    @shiyu1994
    shiyu1994 committed Jun 9, 2021
    Configuration menu
    Copy the full SHA
    0168d2c View commit details
    Browse the repository at this point in the history

  2. fix sync best split from task blocks

    @shiyu1994
    shiyu1994 committed Jun 9, 2021
    Configuration menu
    Copy the full SHA
    0570fe0 View commit details
    Browse the repository at this point in the history

  3. add support for sparse row wise for CUDA

    @shiyu1994
    shiyu1994 committed Jun 9, 2021
    Configuration menu
    Copy the full SHA
    374018c View commit details
    Browse the repository at this point in the history

  4. remove useless code

    @shiyu1994
    shiyu1994 committed Jun 9, 2021
    Configuration menu
    Copy the full SHA
    40c49cc View commit details
    Browse the repository at this point in the history

Commits on Jun 10, 2021

  1. add l2 regression objective

    @shiyu1994
    shiyu1994 committed Jun 10, 2021
    Configuration menu
    Copy the full SHA
    dc41a00 View commit details
    Browse the repository at this point in the history

Commits on Jun 11, 2021

  1. sparse multi value bin enabled for CUDA

    @shiyu1994
    shiyu1994 committed Jun 11, 2021
    Configuration menu
    Copy the full SHA
    bd065b7 View commit details
    Browse the repository at this point in the history

Commits on Jun 16, 2021

  1. fix cuda ranking objective

    @shiyu1994
    shiyu1994 committed Jun 16, 2021
    Configuration menu
    Copy the full SHA
    a5fadfb View commit details
    Browse the repository at this point in the history

  2. support for number of items <= 2048 per query

    Yu Shi committed Jun 16, 2021
    Configuration menu
    Copy the full SHA
    3202b79 View commit details
    Browse the repository at this point in the history

Commits on Jun 23, 2021

  1. speedup histogram construction by interleaving global memory access

    Yu Shi committed Jun 23, 2021
    Configuration menu
    Copy the full SHA
    cd687c9 View commit details
    Browse the repository at this point in the history

Commits on Jun 28, 2021

  1. split optimization

    Yu Shi committed Jun 28, 2021
    Configuration menu
    Copy the full SHA
    320c449 View commit details
    Browse the repository at this point in the history

Commits on Jul 2, 2021

  1. add cuda tree predictor

    Yu Shi committed Jul 2, 2021
    Configuration menu
    Copy the full SHA
    eb1d7fa View commit details
    Browse the repository at this point in the history

Commits on Jul 18, 2021

  1. remove comma

    Yu Shi committed Jul 18, 2021
    Configuration menu
    Copy the full SHA
    dd177f5 View commit details
    Browse the repository at this point in the history

Commits on Jul 19, 2021

  1. refactor objective and score updater

    Yu Shi committed Jul 19, 2021
    Configuration menu
    Copy the full SHA
    ee836d6 View commit details
    Browse the repository at this point in the history

Commits on Jul 21, 2021

  1. before use struct

    Yu Shi committed Jul 21, 2021
    Configuration menu
    Copy the full SHA
    0467fce View commit details
    Browse the repository at this point in the history

  2. use structure for split information

    Yu Shi committed Jul 21, 2021
    Configuration menu
    Copy the full SHA
    f05da3c View commit details
    Browse the repository at this point in the history

  3. use structure for leaf splits

    Yu Shi committed Jul 21, 2021
    Configuration menu
    Copy the full SHA
    400622a View commit details
    Browse the repository at this point in the history

  4. return CUDASplitInfo directly after finding best split

    Yu Shi committed Jul 21, 2021
    Configuration menu
    Copy the full SHA
    d9d3aa9 View commit details
    Browse the repository at this point in the history

Commits on Jul 22, 2021

  1. split with CUDATree directly

    Yu Shi committed Jul 22, 2021
    Configuration menu
    Copy the full SHA
    45cf7a7 View commit details
    Browse the repository at this point in the history

Commits on Jul 26, 2021

  1. use cuda row data in cuda histogram constructor

    Yu Shi committed Jul 26, 2021
    Configuration menu
    Copy the full SHA
    9dea18d View commit details
    Browse the repository at this point in the history

  2. clean src/treelearner/cuda

    Yu Shi committed Jul 26, 2021
    Configuration menu
    Copy the full SHA
    572e2b0 View commit details
    Browse the repository at this point in the history

Commits on Jul 27, 2021

  1. gather shared cuda device functions

    Yu Shi committed Jul 27, 2021
    Configuration menu
    Copy the full SHA
    fe58d4c View commit details
    Browse the repository at this point in the history

  2. put shared CUDA functions into header file

    Yu Shi committed Jul 27, 2021
    Configuration menu
    Copy the full SHA
    dc461dc View commit details
    Browse the repository at this point in the history

  3. change smaller leaf from <= back to < for consistent result with CPU

    Yu Shi committed Jul 27, 2021
    Configuration menu
    Copy the full SHA
    ba565c1 View commit details
    Browse the repository at this point in the history

Commits on Aug 3, 2021

  1. add tree predictor

    Yu Shi committed Aug 3, 2021
    Configuration menu
    Copy the full SHA
    a781ef5 View commit details
    Browse the repository at this point in the history

  2. remove useless cuda_tree_predictor

    Yu Shi committed Aug 3, 2021
    Configuration menu
    Copy the full SHA
    c8a6fab View commit details
    Browse the repository at this point in the history

Commits on Aug 4, 2021

  1. predict on CUDA with pipeline

    Yu Shi committed Aug 4, 2021
    Configuration menu
    Copy the full SHA
    a7504dc View commit details
    Browse the repository at this point in the history

Commits on Aug 9, 2021

  1. add global sort algorithms

    Yu Shi committed Aug 9, 2021
    Configuration menu
    Copy the full SHA
    896d47b View commit details
    Browse the repository at this point in the history

  2. add global argsort for queries with many items in ranking tasks

    Yu Shi committed Aug 9, 2021
    Configuration menu
    Copy the full SHA
    fe6ed74 View commit details
    Browse the repository at this point in the history

Commits on Aug 11, 2021

  1. remove limitation of maximum number of items per query in ranking

    Yu Shi committed Aug 11, 2021
    Configuration menu
    Copy the full SHA
    7808455 View commit details
    Browse the repository at this point in the history

Commits on Aug 16, 2021

  1. add cuda metrics

    Yu Shi committed Aug 16, 2021
    Configuration menu
    Copy the full SHA
    7a0d218 View commit details
    Browse the repository at this point in the history

Commits on Aug 18, 2021

  1. fix CUDA AUC

    Yu Shi committed Aug 18, 2021
    Configuration menu
    Copy the full SHA
    ca42f3b View commit details
    Browse the repository at this point in the history

  2. remove debug code

    Yu Shi committed Aug 18, 2021
    Configuration menu
    Copy the full SHA
    c681102 View commit details
    Browse the repository at this point in the history

Commits on Aug 19, 2021

  1. add regression metrics

    Yu Shi committed Aug 19, 2021
    Configuration menu
    Copy the full SHA
    ea60566 View commit details
    Browse the repository at this point in the history

  2. remove useless file

    Yu Shi committed Aug 19, 2021
    Configuration menu
    Copy the full SHA
    5c84788 View commit details
    Browse the repository at this point in the history

  3. don't use mask in shuffle reduce

    Yu Shi committed Aug 19, 2021
    Configuration menu
    Copy the full SHA
    c2c2407 View commit details
    Browse the repository at this point in the history

Commits on Sep 2, 2021

  1. add more regression objectives

    Yu Shi committed Sep 2, 2021
    Configuration menu
    Copy the full SHA
    b43d367 View commit details
    Browse the repository at this point in the history

Commits on Sep 3, 2021

  1. fix cuda mape loss 

    add cuda xentropy loss
    Yu Shi committed Sep 3, 2021
    Configuration menu
    Copy the full SHA
    951aa37 View commit details
    Browse the repository at this point in the history

  2. use template for different versions of BitonicArgSortDevice

    Yu Shi committed Sep 3, 2021
    Configuration menu
    Copy the full SHA
    b50ce5b View commit details
    Browse the repository at this point in the history

Commits on Sep 6, 2021

  1. add multiclass metrics

    Yu Shi committed Sep 6, 2021
    Configuration menu
    Copy the full SHA
    f51fd70 View commit details
    Browse the repository at this point in the history

Commits on Sep 7, 2021

  1. add ndcg metric

    Yu Shi committed Sep 7, 2021
    Configuration menu
    Copy the full SHA
    35c742d View commit details
    Browse the repository at this point in the history

Commits on Sep 10, 2021

  1. fix cross entropy objectives and metrics

    Yu Shi committed Sep 10, 2021
    Configuration menu
    Copy the full SHA
    510d878 View commit details
    Browse the repository at this point in the history

  2. fix cross entropy and ndcg metrics

    Yu Shi committed Sep 10, 2021
    Configuration menu
    Copy the full SHA
    95f4612 View commit details
    Browse the repository at this point in the history

  3. add support for customized objective in CUDA

    Yu Shi committed Sep 10, 2021
    Configuration menu
    Copy the full SHA
    bb997d0 View commit details
    Browse the repository at this point in the history

  4. complete multiclass ova for CUDA

    Yu Shi committed Sep 10, 2021
    Configuration menu
    Copy the full SHA
    17b78d1 View commit details
    Browse the repository at this point in the history

Commits on Sep 13, 2021

  1. merge master

    Yu Shi committed Sep 13, 2021
    Configuration menu
    Copy the full SHA
    72aa863 View commit details
    Browse the repository at this point in the history

Commits on Oct 24, 2021

  1. remove CUDA_ARCH 6.2 temporarily

    @shiyu1994
    shiyu1994 committed Oct 24, 2021
    Configuration menu
    Copy the full SHA
    c1b8f99 View commit details
    Browse the repository at this point in the history