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

DPU embedding table implementation with DPU multiColumn strategy #21

Open
wants to merge 98 commits into
base: master
Choose a base branch
from
Open

DPU embedding table implementation with DPU multiColumn strategy #21

wants to merge 98 commits into from

Commits on Mar 15, 2022

  1. first version of multi-table

    Niloofar Zarif committed Mar 15, 2022
    Configuration menu
    Copy the full SHA
    2dfece2 View commit details
    Browse the repository at this point in the history

  2. minor changes

    Niloofar Zarif committed Mar 15, 2022
    Configuration menu
    Copy the full SHA
    d675ea3 View commit details
    Browse the repository at this point in the history

Commits on Mar 20, 2022

  1. Added type casting for host code

    @justin-wong-ce
    justin-wong-ce committed Mar 20, 2022
    Configuration menu
    Copy the full SHA
    0862d75 View commit details
    Browse the repository at this point in the history

  2. Added custom Pytorch submodule

    @justin-wong-ce
    justin-wong-ce committed Mar 20, 2022
    Configuration menu
    Copy the full SHA
    f21a0a5 View commit details
    Browse the repository at this point in the history

Commits on Mar 25, 2022

  1. Updated submodule commit ref

    @justin-wong-ce
    justin-wong-ce committed Mar 25, 2022
    Configuration menu
    Copy the full SHA
    8121f5f View commit details
    Browse the repository at this point in the history

Commits on Mar 27, 2022

  1. updated submodules refs

    @justin-wong-ce
    justin-wong-ce committed Mar 27, 2022
    Configuration menu
    Copy the full SHA
    7ab442f View commit details
    Browse the repository at this point in the history

Commits on Mar 30, 2022

  1. added dpu_set_ptr

    Niloofar Zarif committed Mar 30, 2022
    Configuration menu
    Copy the full SHA
    69b4f8d View commit details
    Browse the repository at this point in the history

Commits on Mar 31, 2022

  1. changed lookup() casting

    @justin-wong-ce
    justin-wong-ce committed Mar 31, 2022
    Configuration menu
    Copy the full SHA
    5c8ffc7 View commit details
    Browse the repository at this point in the history

Commits on May 6, 2022

  1. Updated makefile to compile load generator

    @justin-wong-ce
    justin-wong-ce committed May 6, 2022
    Configuration menu
    Copy the full SHA
    dddafd4 View commit details
    Browse the repository at this point in the history

Commits on May 11, 2022

  1. emb_host runs with ./run.sh -br toy, still seg faults

    @justin-wong-ce
    justin-wong-ce committed May 11, 2022
    Configuration menu
    Copy the full SHA
    3eea8bb View commit details
    Browse the repository at this point in the history

Commits on May 17, 2022

  1. makefile and load_generator runs

    @justin-wong-ce
    justin-wong-ce committed May 17, 2022
    Configuration menu
    Copy the full SHA
    8e77888 View commit details
    Browse the repository at this point in the history

Commits on May 19, 2022

  1. pushing latest version which stalls

    Niloofar Zarif committed May 19, 2022
    Configuration menu
    Copy the full SHA
    dbaadca View commit details
    Browse the repository at this point in the history

Commits on May 20, 2022

  1. updated commit refs, updated ./upmem structure

    @justin-wong-ce
    justin-wong-ce committed May 20, 2022
    Configuration menu
    Copy the full SHA
    223af36 View commit details
    Browse the repository at this point in the history

  2. Merge branch 'loadgen' of https://github.com/UBC-ECE-Sasha/PIM-Embedd… 

    …ing-Lookup into loadgen
    @justin-wong-ce
    justin-wong-ce committed May 20, 2022
    Configuration menu
    Copy the full SHA
    3a419dd View commit details
    Browse the repository at this point in the history

Commits on Jun 1, 2022

  1. profiling dpu vs cpu

    Niloofar Zarif committed Jun 1, 2022
    Configuration menu
    Copy the full SHA
    a53c196 View commit details
    Browse the repository at this point in the history

  2. adding new dlrm

    Niloofar Zarif committed Jun 1, 2022
    Configuration menu
    Copy the full SHA
    9d67b74 View commit details
    Browse the repository at this point in the history

Commits on Jun 29, 2022

  1. added dpu_per_rank flexibility

    Niloofar Zarif committed Jun 29, 2022
    Configuration menu
    Copy the full SHA
    09844e4 View commit details
    Browse the repository at this point in the history

Commits on Jul 1, 2022

  1. pointing to own submodule

    @SylvanBrocard
    SylvanBrocard committed Jul 1, 2022
    Configuration menu
    Copy the full SHA
    274b0b0 View commit details
    Browse the repository at this point in the history

  2. added clangd config file

    @SylvanBrocard
    SylvanBrocard committed Jul 1, 2022
    Configuration menu
    Copy the full SHA
    ce57e2c View commit details
    Browse the repository at this point in the history

  3. updated for includes

    @SylvanBrocard
    SylvanBrocard committed Jul 1, 2022
    Configuration menu
    Copy the full SHA
    0006b08 View commit details
    Browse the repository at this point in the history

Commits on Jul 4, 2022

  1. fixed validation function

    @SylvanBrocard
    SylvanBrocard committed Jul 4, 2022
    Configuration menu
    Copy the full SHA
    e3ef711 View commit details
    Browse the repository at this point in the history

  2. code refacto + naming

    Dimitri Gerin committed Jul 4, 2022
    Configuration menu
    Copy the full SHA
    ad83789 View commit details
    Browse the repository at this point in the history

  3. streamlined kernel

    @SylvanBrocard
    SylvanBrocard committed Jul 4, 2022
    Configuration menu
    Copy the full SHA
    dcee4d3 View commit details
    Browse the repository at this point in the history

  4. fix synthetic_populate random generation

    Dimitri Gerin committed Jul 4, 2022
    Configuration menu
    Copy the full SHA
    acfafb9 View commit details
    Browse the repository at this point in the history

  5. codebase refactoring

    Dimitri Gerin committed Jul 4, 2022
    Configuration menu
    Copy the full SHA
    fa10a74 View commit details
    Browse the repository at this point in the history

Commits on Jul 5, 2022

  1. fix WARNINGS

    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    01caff0 View commit details
    Browse the repository at this point in the history

  2. small scenario

    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    d57fa3c View commit details
    Browse the repository at this point in the history

  3. small scenario

    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    4689c68 View commit details
    Browse the repository at this point in the history

  4. add tracing facilities to Makefile

    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    96059d3 View commit details
    Browse the repository at this point in the history

  5. comments & naming

    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    af67194 View commit details
    Browse the repository at this point in the history

  6. fix MAX_NR_BATCHES

    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    0599741 View commit details
    Browse the repository at this point in the history

  7. redo populate mram

    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    3687488 View commit details
    Browse the repository at this point in the history

  8. populate fix

    @SylvanBrocard
    SylvanBrocard committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    b88ef69 View commit details
    Browse the repository at this point in the history

  9. moved post-processing outside callback

    @SylvanBrocard
    SylvanBrocard committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    4a81d72 View commit details
    Browse the repository at this point in the history

  10. removed mem_reset

    @SylvanBrocard
    SylvanBrocard committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    f5d4f72 View commit details
    Browse the repository at this point in the history

  11. prevent overflow in synthetic tests

    @SylvanBrocard
    SylvanBrocard committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    98ec3d9 View commit details
    Browse the repository at this point in the history

  12. code refacto for emb_tables 

    # Conflicts:
#	upmem/src/synthetic_dataset_embedding.c
    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    4b06262 View commit details
    Browse the repository at this point in the history

  13. add constant def && profiling app func

    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    fcfcec3 View commit details
    Browse the repository at this point in the history

  14. fixed post-processing

    @SylvanBrocard
    SylvanBrocard committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    4d5ee8c View commit details
    Browse the repository at this point in the history

  15. fix missing NR_TASKLETS && code cleaning

    Dimitri Gerin committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    ab4f839 View commit details
    Browse the repository at this point in the history

  16. preventing overflow in tests

    @SylvanBrocard
    SylvanBrocard committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    eddb382 View commit details
    Browse the repository at this point in the history

  17. explicit loop control

    @SylvanBrocard
    SylvanBrocard committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    8f98712 View commit details
    Browse the repository at this point in the history

  18. explicit for loop

    @SylvanBrocard
    SylvanBrocard committed Jul 5, 2022
    Configuration menu
    Copy the full SHA
    ae53265 View commit details
    Browse the repository at this point in the history

Commits on Jul 6, 2022

  1. use uint64_t for dimentional variables && naming && result buffer all… 

    …ocation refacto
    Dimitri Gerin committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    37f0a30 View commit details
    Browse the repository at this point in the history

  2. restored handling for large nr of indices

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    56818db View commit details
    Browse the repository at this point in the history

  3. Merge previous commits

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    94bb9a2 View commit details
    Browse the repository at this point in the history

  4. commented debug code

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    52a5a37 View commit details
    Browse the repository at this point in the history

  5. wrong assert

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    c1e4584 View commit details
    Browse the repository at this point in the history

  6. format && replact uint32_t by uint64_t

    Dimitri Gerin committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    0907925 View commit details
    Browse the repository at this point in the history

  7. preparation for async

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    1f6856b View commit details
    Browse the repository at this point in the history

  8. imports cleanup

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    eefea21 View commit details
    Browse the repository at this point in the history

  9. unused parameter

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    e76261e View commit details
    Browse the repository at this point in the history

  10. fixed time counter

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    25dd8ba View commit details
    Browse the repository at this point in the history

  11. added perf counters + experiment

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    986c409 View commit details
    Browse the repository at this point in the history

  12. BUG & PERF fix : separate alloc_dpu_results

    Dimitri Gerin committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    22ab4fd View commit details
    Browse the repository at this point in the history

  13. preparation for async

    @SylvanBrocard
    SylvanBrocard authored and Dimitri Gerin committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    190dd61 View commit details
    Browse the repository at this point in the history

  14. refacto : dont use precompiler constants

    Dimitri Gerin committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    4197901 View commit details
    Browse the repository at this point in the history

  15. 32 bits int in kernel

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    377c960 View commit details
    Browse the repository at this point in the history

  16. formatting

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    0de53a0 View commit details
    Browse the repository at this point in the history

  17. now supports >512 batches

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    11c891a View commit details
    Browse the repository at this point in the history

  18. refacto : separates indices, offsets and input info allocation

    Dimitri Gerin committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    946f44c View commit details
    Browse the repository at this point in the history

  19. param : switch to original

    Dimitri Gerin committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    4b6a46c View commit details
    Browse the repository at this point in the history

  20. fixed

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    c6426db View commit details
    Browse the repository at this point in the history

  21. put counters in ifdef blocks

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    973ae76 View commit details
    Browse the repository at this point in the history

  22. updated clangd

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    2dd274c View commit details
    Browse the repository at this point in the history

  23. imports cleanup

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    24b382d View commit details
    Browse the repository at this point in the history

  24. unused parameter

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    3fd8ffe View commit details
    Browse the repository at this point in the history

  25. fixed time counter

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    b8a5b02 View commit details
    Browse the repository at this point in the history

  26. added perf counters + experiment

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    de8acbb View commit details
    Browse the repository at this point in the history

  27. 32 bits int in kernel

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    9ac37f3 View commit details
    Browse the repository at this point in the history

  28. formatting

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    eaca4e2 View commit details
    Browse the repository at this point in the history

  29. now supports >512 batches

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    5fc81ac View commit details
    Browse the repository at this point in the history

  30. put counters in ifdef blocks

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    390254b View commit details
    Browse the repository at this point in the history

  31. updated clangd

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    0cf34fb View commit details
    Browse the repository at this point in the history

  32. measuring verification time

    @SylvanBrocard
    SylvanBrocard committed Jul 6, 2022
    Configuration menu
    Copy the full SHA
    4cafc19 View commit details
    Browse the repository at this point in the history

Commits on Jul 7, 2022

  1. pipeline stage 1 [build_synthetic_input_data -> synthetic_inference]

    Dimitri Gerin committed Jul 7, 2022
    Configuration menu
    Copy the full SHA
    b16b4f4 View commit details
    Browse the repository at this point in the history

  2. Merge remote-tracking branch 'origin/sbrocard/dpu_profiling' into dge… 

    …rin/pipeline

# Conflicts:
#	upmem/Makefile
#	upmem/src/embedding.c
#	upmem/src/synthetic_dataset_embedding.c
    Dimitri Gerin committed Jul 7, 2022
    Configuration menu
    Copy the full SHA
    06d3d35 View commit details
    Browse the repository at this point in the history

  3. few merge fix

    Dimitri Gerin committed Jul 7, 2022
    Configuration menu
    Copy the full SHA
    a407ff6 View commit details
    Browse the repository at this point in the history

  4. post processing rank callback

    Dimitri Gerin committed Jul 7, 2022
    Configuration menu
    Copy the full SHA
    94daeba View commit details
    Browse the repository at this point in the history

  5. bug fix

    @SylvanBrocard
    SylvanBrocard committed Jul 7, 2022
    Configuration menu
    Copy the full SHA
    08d15c4 View commit details
    Browse the repository at this point in the history

Commits on Jul 8, 2022

  1. Merge branch 'merge_tmp' into pipelinemerge 

    # Conflicts:
#	upmem/Makefile
#	upmem/src/dpu/dpu_embedding.c
#	upmem/src/synthetic_dataset_embedding.c
    Dimitri Gerin committed Jul 8, 2022
    Configuration menu
    Copy the full SHA
    4224a9f View commit details
    Browse the repository at this point in the history

  2. Merge commit '94daeba2e6264733e4c46a2ccd9d2704bdb6b8ba' into dgerin/p… 

    …ipeline

# Conflicts:
#	upmem/Makefile
#	upmem/src/embedding.c
#	upmem/src/synthetic_dataset_embedding.c
    Dimitri Gerin committed Jul 8, 2022
    Configuration menu
    Copy the full SHA
    5eda412 View commit details
    Browse the repository at this point in the history

  3. BUGGY rank_mapping not functional

    Dimitri Gerin committed Jul 8, 2022
    Configuration menu
    Copy the full SHA
    c4fe841 View commit details
    Browse the repository at this point in the history

  4. merge with host pipelining (synthetic_data->inference->gather_dpu_res… 

    …ults(callback)) (#1)

* preparation for async

* imports cleanup

* unused parameter

* fixed time counter

* added perf counters + experiment

* BUG & PERF fix : separate alloc_dpu_results

* preparation for async

* refacto : dont use precompiler constants

* 32 bits int in kernel

* formatting

* now supports >512 batches

* refacto : separates indices, offsets and input info allocation

* param : switch to original

* fixed

* put counters in ifdef blocks

* updated clangd

* imports cleanup

* unused parameter

* fixed time counter

* added perf counters + experiment

* 32 bits int in kernel

* formatting

* now supports >512 batches

* put counters in ifdef blocks

* updated clangd

* measuring verification time

* pipeline stage 1 [build_synthetic_input_data -> synthetic_inference]

* few merge fix

* bug fix

Co-authored-by: SylvanBrocard <[email protected]>
    @dgerinmem @SylvanBrocard
    dgerinmem and SylvanBrocard authored Jul 8, 2022
    Configuration menu
    Copy the full SHA
    df2e350 View commit details
    Browse the repository at this point in the history

  5. fix inconsistency : release mode host side compilation 

    bugfix time measurement
use 10 re-run for DPU
    Dimitri Gerin committed Jul 8, 2022
    Configuration menu
    Copy the full SHA
    e2c0cb8 View commit details
    Browse the repository at this point in the history

Commits on Jul 11, 2022

  1. Merge remote-tracking branch 'origin/upmem_internal' into pipeline 

    # Conflicts:
#	upmem/Makefile
#	upmem/include/emb_types.h
#	upmem/include/embedding.h
#	upmem/src/embedding.c
#	upmem/src/synthetic_dataset_embedding.c
    Dimitri Gerin committed Jul 11, 2022
    Configuration menu
    Copy the full SHA
    817b8ab View commit details
    Browse the repository at this point in the history

Commits on Jul 13, 2022

  1. flexible rank/embedding mapping feature

    Dimitri Gerin committed Jul 13, 2022
    Configuration menu
    Copy the full SHA
    140d845 View commit details
    Browse the repository at this point in the history

  2. clock_gettime() fix clock type : use Wall time

    Dimitri Gerin committed Jul 13, 2022
    Configuration menu
    Copy the full SHA
    ab57898 View commit details
    Browse the repository at this point in the history

Commits on Jul 20, 2022

  1. bugfix alloc_dpus 

    feature : RAND_INPUT_SIZE=1 -> synthetic input indices generation
                               with fixed batch size, random Pooling Factor (num indices per lookup)
add PERF.md for perf log
    Dimitri Gerin committed Jul 20, 2022
    Configuration menu
    Copy the full SHA
    1ad486d View commit details
    Browse the repository at this point in the history

Commits on Jul 28, 2022

  1. multicol && rm2 perf

    Dimitri Gerin committed Jul 28, 2022
    Configuration menu
    Copy the full SHA
    6c3cc4f View commit details
    Browse the repository at this point in the history

Commits on Aug 23, 2022

  1. refacto separate CPU look and CPU/DPU check 

            separate testbench and embedding lib
        clean names
        update README
    Dimitri Gerin committed Aug 23, 2022
    Configuration menu
    Copy the full SHA
    9f8b654 View commit details
    Browse the repository at this point in the history

Commits on Sep 29, 2022

  1. [bugfix] nr_ranks used before read actual number of ranks

    Dimitri Gerin committed Sep 29, 2022
    Configuration menu
    Copy the full SHA
    72bd435 View commit details
    Browse the repository at this point in the history

  2. [app] code cleaning & add comments about mapping

    Dimitri Gerin committed Sep 29, 2022
    Configuration menu
    Copy the full SHA
    04b4b60 View commit details
    Browse the repository at this point in the history

  3. [bugfix] memory leak missing free()

    Dimitri Gerin committed Sep 29, 2022
    Configuration menu
    Copy the full SHA
    a705c77 View commit details
    Browse the repository at this point in the history

  4. [bugfix] bad loop informations cause memory fault

    Dimitri Gerin committed Sep 29, 2022
    Configuration menu
    Copy the full SHA
    d13cf0a View commit details
    Browse the repository at this point in the history

  5. [app] use DPU ASYNC jobs

    Dimitri Gerin committed Sep 29, 2022
    Configuration menu
    Copy the full SHA
    d87cb6c View commit details
    Browse the repository at this point in the history

  6. [app] move lenght structure allocation out of critical func

    Dimitri Gerin committed Sep 29, 2022
    Configuration menu
    Copy the full SHA
    9028bbd View commit details
    Browse the repository at this point in the history

  7. [profiling] fix APP function profiling

    Dimitri Gerin committed Sep 29, 2022
    Configuration menu
    Copy the full SHA
    766531c View commit details
    Browse the repository at this point in the history