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

Bfloat16 #1709

Draft
wants to merge 48 commits into
base: develop
Choose a base branch
from
Draft

Bfloat16 #1709

wants to merge 48 commits into from

Commits on Sep 29, 2023

  1. only can compile cuda/omp

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    c0f2d9a View commit details
    Browse the repository at this point in the history

  2. next_precision to itself when complex only float, double add empty co… 

    …nditional
    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    ac1dbb2 View commit details
    Browse the repository at this point in the history

  3. can compile with cuda/omp/ref (without test)

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    7bffc78 View commit details
    Browse the repository at this point in the history

  4. compile for cuda/sycl/test/mpi (hip needs trick)

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    c517644 View commit details
    Browse the repository at this point in the history

  5. hip finally

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    6b5a4ec View commit details
    Browse the repository at this point in the history

  6. fix the narrow issue and atomic support

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    add37ab View commit details
    Browse the repository at this point in the history

  7. fixed more error

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    2f53fce View commit details
    Browse the repository at this point in the history

  8. fix the op order and gdb 

    Co-authored-by: Marcel Koch <[email protected]>
    @yhmtsai @MarcelKoch
    yhmtsai and MarcelKoch committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    cdc4d6b View commit details
    Browse the repository at this point in the history

  9. add the rand template not_implemented

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    c1c1551 View commit details
    Browse the repository at this point in the history

  10. this version can compile/run complex<half> on cuda114 

    stack size can not be determined statically issue
some tests are failed
    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    3f11657 View commit details
    Browse the repository at this point in the history

  11. does not work for the other executor

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    6b9546b View commit details
    Browse the repository at this point in the history

  12. fix complex issue and sqrt issue 

    sqrt need to be global namespace to live with hip sqrt
    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    384371b View commit details
    Browse the repository at this point in the history

  13. try fix the compilation issue from MSVC and MacOS 

    It seems to use complex version even using half only
    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    c04d7d0 View commit details
    Browse the repository at this point in the history

  14. move the half to public and use sycl::half for dpcpp

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    efb9aea View commit details
    Browse the repository at this point in the history

  15. limit the next precision in test and benchmark 

    next_precision<double> = float not half
    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    9480b50 View commit details
    Browse the repository at this point in the history

  16. allow disable half operation

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    9625e11 View commit details
    Browse the repository at this point in the history

  17. fix macro

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    620aef5 View commit details
    Browse the repository at this point in the history

  18. clean and refine the code

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    06a6f46 View commit details
    Browse the repository at this point in the history

  19. move half.hpp out of type.hpp

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    814dca4 View commit details
    Browse the repository at this point in the history

  20. enable half for testing

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    54755b4 View commit details
    Browse the repository at this point in the history

  21. __habs is added in cuda10.2 

    create_empty for its own type
    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    cc3a7d5 View commit details
    Browse the repository at this point in the history

  22. fix nullptr and missing instantiation. 

    sycl::half has different rule in conv and full operator after 5.7
    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    7af58c9 View commit details
    Browse the repository at this point in the history

  23. fix missing device_type and ptr_param

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    35a47fa View commit details
    Browse the repository at this point in the history

  24. update rounding

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    faea48d View commit details
    Browse the repository at this point in the history

  25. do not use distribution with half

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    f1c1181 View commit details
    Browse the repository at this point in the history

  26. WIP fix half of failed test

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    d450dc5 View commit details
    Browse the repository at this point in the history

  27. fix/skip half test and fix numeric_limit on device

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    e83d53d View commit details
    Browse the repository at this point in the history

  28. mkl csr does not support half

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    2d7f4cb View commit details
    Browse the repository at this point in the history

  29. add half to batch_vector

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    b04c993 View commit details
    Browse the repository at this point in the history

  30. fix hip thrust complex op, avoid const in nvhpc, reduce job in windows

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    10bb4ae View commit details
    Browse the repository at this point in the history

  31. fix nvc++ atomic, dpcpp half 

    jacobi need to use value_type generator
    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    bf25352 View commit details
    Browse the repository at this point in the history

  32. make half test optional

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    cf9c218 View commit details
    Browse the repository at this point in the history

  33. nvhpc optimization/computation error workaround

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    1e4b68b View commit details
    Browse the repository at this point in the history

  34. disable mpi half test

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    c6bbf20 View commit details
    Browse the repository at this point in the history

  35. some math func is not defined if nvhpc is for host

    @yhmtsai
    yhmtsai committed Sep 29, 2023
    Configuration menu
    Copy the full SHA
    18d323e View commit details
    Browse the repository at this point in the history

Commits on Sep 30, 2023

  1. add half spmv benchmark (with cusparse for cuda)

    @yhmtsai
    yhmtsai committed Sep 30, 2023
    Configuration menu
    Copy the full SHA
    8b27e3c View commit details
    Browse the repository at this point in the history

Commits on Oct 1, 2023

  1. add bfloat16

    @yhmtsai
    yhmtsai committed Oct 1, 2023
    Configuration menu
    Copy the full SHA
    4eb3b53 View commit details
    Browse the repository at this point in the history

  2. this can be compiled after cuda arch 80

    @yhmtsai
    yhmtsai committed Oct 1, 2023
    Configuration menu
    Copy the full SHA
    f3913a2 View commit details
    Browse the repository at this point in the history

  3. some trick for cuda arch < 8.0

    @yhmtsai
    yhmtsai committed Oct 1, 2023
    Configuration menu
    Copy the full SHA
    d87dc17 View commit details
    Browse the repository at this point in the history

  4. fix the missing type conversion

    @yhmtsai
    yhmtsai committed Oct 1, 2023
    Configuration menu
    Copy the full SHA
    5f48681 View commit details
    Browse the repository at this point in the history

  5. enable bfloat16 test and fix/skip

    @yhmtsai
    yhmtsai committed Oct 1, 2023
    Configuration menu
    Copy the full SHA
    316775f View commit details
    Browse the repository at this point in the history

Commits on Oct 2, 2023

  1. fix missing type, use bfloat16 op from cudaarch80

    @yhmtsai
    yhmtsai committed Oct 2, 2023
    Configuration menu
    Copy the full SHA
    4f0d12a View commit details
    Browse the repository at this point in the history

  2. fix hip 

    hip does not support float -> bfloat16 implicit conversion
bfloat a = float() does not work
    @yhmtsai
    yhmtsai committed Oct 2, 2023
    Configuration menu
    Copy the full SHA
    8364204 View commit details
    Browse the repository at this point in the history

  3. fix dpcpp

    @yhmtsai
    yhmtsai committed Oct 2, 2023
    Configuration menu
    Copy the full SHA
    95ac5ea View commit details
    Browse the repository at this point in the history

  4. fix nvhpc

    @yhmtsai
    yhmtsai committed Oct 2, 2023
    Configuration menu
    Copy the full SHA
    653cd36 View commit details
    Browse the repository at this point in the history

Commits on Oct 3, 2023

  1. add the casting

    @yhmtsai
    yhmtsai committed Oct 3, 2023
    Configuration menu
    Copy the full SHA
    50cf5b2 View commit details
    Browse the repository at this point in the history

Commits on Oct 4, 2023

  1. use float as the bridge between bfloat16 and half

    @yhmtsai
    yhmtsai committed Oct 4, 2023
    Configuration menu
    Copy the full SHA
    51ab0b0 View commit details
    Browse the repository at this point in the history

  2. fix ell accessor type

    @yhmtsai
    yhmtsai committed Oct 4, 2023
    Configuration menu
    Copy the full SHA
    b531ba3 View commit details
    Browse the repository at this point in the history