GitLab

[Performance] Perform to_block on the GPU when the dataloader is created with a GPU `device`. (#3016)

* add output device for dataloading

* Update dataloader

* Get sampler device from dataloader

* Fix line length

* Update examples

* Fix to_block GPU for empty relation types

* Handle the case where the DistGraph has None for the underlying graph
Co-authored-by: Da Zheng <zhengda1936@gmail.com>

* fix gatedgcn

* fix lint

* Split from NCCL PR

* Fix type in comment

* Expand documentation for sparse_all_to_all_push

* Restore previous behavior in example

* Re-work optimizer to use NCCL based on gradient location

* Allow for running with embedding on CPU but using NCCL for gradient exchange

* Optimize single partition case

* Fix pylint errors

* Add missing include

* fix gradient indexing

* Fix line continuation

* Migrate 'first_step'

* Skip tests without enough GPUs to run NCCL

* Improve empty tensor handling for pytorch 1.5

* Fix indentation

* Allow multiple NCCL communicator to coexist

* Improve handling of empty message

* Update python/dgl/nn/pytorch/sparse_emb.py
Co-authored-by: xiang song(charlie.song) <classicxsong@gmail.com>

* Update python/dgl/nn/pytorch/sparse_emb.py
Co-authored-by: xiang song(charlie.song) <classicxsong@gmail.com>

* Keepy empty tensor dimensionaless

* th.empty -> th.tensor

* Preserve shape for empty non-zero dimension tensors

* Use shared state, when embedding is shared

* Add support for gathering an embedding

* Fix typo

* Fix more typos

* Fix backend call

* Use NodeDataLoader to take advantage of ddp

* Update training script to share memory

* Only squeeze last dimension

* Better handle empty message

* Keep embedding on the target device GPU if dgl_sparse if false in RGCN example

* Fix typo in comment

* Add asserts

* Improve documentation in example
Co-authored-by: xiang song(charlie.song) <classicxsong@gmail.com>

* Make multigpu graphsage use whole datset

* Specify queeze dimension

* Remove squeeze dimension
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Update

* Add files via upload

* Add files via upload

* Add files via upload

* Add files via upload

* Update

* Update

* Add files via upload

* Add files via upload

* Update

* Lint

* Add files via upload

* Lint

* Update

* Update

* Update

* Update

* Update

* Lint Fix

* Lint
Co-authored-by: Ubuntu <ubuntu@ip-172-31-12-161.us-west-2.compute.internal>
Co-authored-by: Jinjing Zhou <VoVAllen@users.noreply.github.com>
Co-authored-by: xiang song(charlie.song) <classicxsong@gmail.com>

* Added a special implementation for MetapathBasedRandomWalkStep for the Uniform randomwalk case

* Fixed all linting issues

* add comment
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>
Co-authored-by: Da Zheng <zhengda1936@gmail.com>

* fix compatibility

* fix

* lint

* multi-GPU node classification tutorial

* Update 2_node_classification.py

* fixes

* elaborate a bit more

* address comments

* address comments

* commit hetgnn

* Update README.md

* Delete examples/pytorch/hetgnn directory

* Update README.md

* Update README.md

* Update README.md

* Revert "Update README.md"

This reverts commit 14cdb3da76c58665ff879eae17ecf9eb28b66cce.

* Update README.md

* Add files via upload

* Update README.md

* Update README.md

* reorg and expose dgl.geometry

* fix lint

* fix test

* fix

* [CPU Parallel] implementation of parallel_for with grain size

* Create parallel_for.h

* Update parallel_for.h
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>

Co-authored-by: zhjwy9343 <6593865@qq.com>

* [Doc] modify the dimension of weight to Numpy broadcasting rule

* [Doc] modify the user defined reduce function
Co-authored-by: zhjwy9343 <6593865@qq.com>

Co-authored-by: zhjwy9343 <6593865@qq.com>
Co-authored-by: Jinjing Zhou <VoVAllen@users.noreply.github.com>

* add init version.

* fix build.

* fix format.

* fix.

* fix.

* fix format.

* update README.

* avoid running CI on distributed training tutorials.

* Update tutorials/dist/1_node_classification.py
Co-authored-by: xiang song(charlie.song) <classicxsong@gmail.com>

* fix.
Co-authored-by: xiang song(charlie.song) <classicxsong@gmail.com>

* add

* fix

* set default

* fix

* try master

* try fix

* try

* fix

* 111

* fix

* fix

* update

* ccc

* try

* fix

* fix

* try new machine

* fix

* fix

* fix

* Revert "fix"

This reverts commit e716d66b046f92fe7ae368947a51a036a7a3188a.

* try

* more parrallel

* use k8s for all

* fix name

* try not specify instance type

* ci

* use one yaml

* Revert "use one yaml"

This reverts commit 717d8d852be39fbf2e2e45f9f224aa97907c372c.

* add timeout

* fix permission

* mount efs

* print

* fix pvc

* fix

* restrict num of gpu instances

* check

* fix

* fix

* make to_heterogeneous and to_homogeneous invertible

* docstring

* oops
Co-authored-by: Jinjing Zhou <VoVAllen@users.noreply.github.com>

* tahin

* readme

* readme

* readme

* readme

* readme

* readme

* main

* main

* new_line

* update
Co-authored-by: zhjwy9343 <6593865@qq.com>

* SpMM for heterograph

* C APIs SDDMM heterograph

* passes initial result

* renamed eid with nid

* aggregation on same ntype for multiple etypes

* fix link check failure

* lint check part 2

* lint check part 3

* Fixed SpMMCmpCsr Min op

* added mem references

* fixed fill(Max/Min), added const

* removed newline

* brought back docstring
Co-authored-by: Israt Nisa <nisisrat@amazon.com>
Co-authored-by: Da Zheng <zhengda1936@gmail.com>

* correct the matrix multiplication symbols

* modified dimensionality of linear
Co-authored-by: Minjie Wang <wmjlyjemaine@gmail.com>

Co-authored-by: Zheng <dzzhen@3c22fba32af5.ant.amazon.com>

* fix regression ram info

* fix

Co-authored-by: xiang song(charlie.song) <classicxsong@gmail.com>

Co-authored-by: Minjie Wang <wmjlyjemaine@gmail.com>

Co-authored-by: Minjie Wang <wmjlyjemaine@gmail.com>

corrected docstring, the model uses two R-GCN layers
Co-authored-by: Minjie Wang <wmjlyjemaine@gmail.com>

* Enable C++ tests on GPU

* Add missing '--runtime nvidia'

* update

* update

* update

* update

* lint

* lint

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* lint

* update

* clone

* update

* update

* update

* update

* replace idarray with ndarray

* refactor cpp part

* refactor python part

* debug

* refactor interface

* test and doc

* lint and test

* lint

* fix

* fix

* fix

* const

* doc

* fix

* fix

* fix

* fix

* fix & doc

* fix

* fix

* fix

* fix

* fix

* fix

* update
Co-authored-by: Minjie Wang <wmjlyjemaine@gmail.com>

Fix a typo in the tutorial for "Introduction of Neighbor Sampling for GNN Training".
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>

* Fix #2808

* fix

* fix

* fix

* fix

* fix

* fix
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>

* Update

* Update

* Update tutorials/multi/1_graph_classification.py
Co-authored-by: Tong He <hetong007@gmail.com>

* Update

* Update

* Update
Co-authored-by: Ubuntu <ubuntu@ip-172-31-4-21.us-west-2.compute.internal>
Co-authored-by: Tong He <hetong007@gmail.com>

* speed up random number generation.

* fix lint.

* Fix

* fix.
Co-authored-by: Zheng <dzzhen@3c22fba32af5.ant.amazon.com>

* add two fraud dataset

* add comments and modify to heterogenous graph

* modify the node name based on datasets

* add assertion message
Co-authored-by: zhjwy9343 <6593865@qq.com>

* [Feature] fix #2876, add two fake news datasets

* add name assertion

* add assertion message
Co-authored-by: zhjwy9343 <6593865@qq.com>