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

Liubo/enable srf avx2 bf16 fp16 #20486

Merged
merged 17 commits into from
Jan 8, 2024
Merged

Liubo/enable srf avx2 bf16 fp16 #20486

merged 17 commits into from
Jan 8, 2024

Commits on Jan 4, 2024

  1. fix avx2 fp16 load_store_emitters issue

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    4bbd204 View commit details
    Browse the repository at this point in the history

  2. add eltwise_node bf16 support, and brgconv_avx2 for common cpu node D… 

    …efaultImplPriority
    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    d375bd5 View commit details
    Browse the repository at this point in the history

  3. MVN node support avx2_bf16

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    cfd846d View commit details
    Browse the repository at this point in the history

  4. limit brgconv support of avx2 to avx2_vnni_2, and enable s8s8 conv on… 

    … avx2_vnni_2
    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    6ab2097 View commit details
    Browse the repository at this point in the history

  5. fix avx2 bf16 memory free issue of Eltwise node

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    e44e5d3 View commit details
    Browse the repository at this point in the history

  6. re-add avx2_vnni_2 configure as temp commit been dropped

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    6565690 View commit details
    Browse the repository at this point in the history

  7. W.A bf16/fp16 of gather jit impl and bf16/fp16 weightsDecompression o… 

    …f fullyconnected not supported on avx2_vnni_2
    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    f4ace00 View commit details
    Browse the repository at this point in the history

  8. fix elementwise functional test 'primTypeCheck' fail error, extend 'h… 

    …asHardwareSupport' function on avx2_vnni_2
    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    d1494a2 View commit details
    Browse the repository at this point in the history

  9. use avx2_vnni_2 new convert instructions

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    31bd841 View commit details
    Browse the repository at this point in the history

  10. fix simple_if testcase failed issue

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    0b753ba View commit details
    Browse the repository at this point in the history

  11. remove gather Node WA, as gpt-j-6b AC issue has been fixed

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    eb9aa08 View commit details
    Browse the repository at this point in the history

  12. fix rebase conflict with new api

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    304d8e4 View commit details
    Browse the repository at this point in the history

  13. change default inferencePrecision from bf16 to f32 on avx2_vnni_2

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    f864ce2 View commit details
    Browse the repository at this point in the history

  14. skip prepare_table() in jit_uni_vcvtneps2bf16 for avx2_vnni_2

    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    39c3d87 View commit details
    Browse the repository at this point in the history

  15. loose fp16 support limitation of jit_load_store_emitters to avx512_co… 

    …re and avx2
    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    d6400d1 View commit details
    Browse the repository at this point in the history

  16. Apply suggestions from code review: remove 'ov::with_cpu_x86_avx2_vnn… 

    …i_2' api, fix store_num and OPENVINO_THROW issue, keep brgconv related types only for new platform priorities list, update WeightsDecompressionImpl of fullyconnected Node, replace mayiuse() by hasHardwareSupport() for Precision check
    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    6a94fe2 View commit details
    Browse the repository at this point in the history

  17. hasHardwareSupport replacements, Conv DefaultImplPriority erase condi… 

    …ation for all brgconv
    @liubo-intel
    liubo-intel committed Jan 4, 2024
    Configuration menu
    Copy the full SHA
    104b6e2 View commit details
    Browse the repository at this point in the history