[Feature] ARGO: an easy-to-use runtime to improve GNN training performance on multi-core processors #7003

jasonlin316 · 2024-01-23T20:52:51Z

Description

The GNN training performance on multi-core processors is limited as the current design cannot scale well. We propose a runtime system named ARGO that can improve the scalability of GNN training on multi-core processors. On a CPU platform where the original program can only scale to 16 cores (meaning that no performance improvement is achieved if more than 16 cores are applied), ARGO can further scale the design up to 64 cores, achieving up to 4.3x speedup compared to the original design without ARGO.

Checklist

Please feel free to remove inapplicable items for your PR.

The PR title starts with [$CATEGORY] (such as [NN], [Model], [Doc], [Feature]])
I've leverage the tools to beautify the python and c++ code.
The PR is complete and small, read the Google eng practice (CL equals to PR) to understand more about small PR. In DGL, we consider PRs with less than 200 lines of core code change are small (example, test and documentation could be exempted).
All changes have test coverage
Code is well-documented
To the best of my knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change
If the PR is for a new model/paper, I've updated the example index here.

Changes

dgl-bot · 2024-01-23T20:53:22Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-01-23T20:53:34Z

Commit ID: 8bfa024

Build ID: 1

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

anko-intel

Please adopt the change to the fact that now it is integral part of DGL example and not external repository.

examples/README.md

examples/pytorch/ARGO/argo.py

examples/pytorch/ARGO/README.md

examples/pytorch/ARGO/ogb_example.py

examples/pytorch/ARGO/ogb_example_ARGO.py

examples/pytorch/ARGO/ogb_example.py

examples/README.md

dgl-bot · 2024-02-05T23:57:58Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-05T23:58:11Z

Commit ID: 24bce24f6d13bf4eaf31a1275bae337c491d6195

Build ID: 2

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

dgl-bot · 2024-02-06T00:04:57Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-06T00:05:09Z

Commit ID: aa3318628bf1e0993d2f88011fa3a3568e52b36b

Build ID: 3

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

dgl-bot · 2024-02-06T00:10:02Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-06T00:10:14Z

Commit ID: 02c3264751590fee73a105c4f8615cf01dc332b8

Build ID: 4

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

dgl-bot · 2024-02-06T00:10:22Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-06T00:10:34Z

Commit ID: 20282490bf28fa6a3e3e3f86ed3cf400ad844a6b

Build ID: 5

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

dgl-bot · 2024-02-06T00:15:58Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-06T00:16:10Z

Commit ID: c8e77a58093dc0d5777974473a0a7c9c60f21219

Build ID: 6

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

dgl-bot · 2024-02-06T00:19:32Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-06T00:19:44Z

Commit ID: 4306493eafb43b7565521d94dcdc596f6f9af2c7

Build ID: 7

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

dgl-bot · 2024-02-06T00:21:17Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-06T00:21:30Z

Commit ID: f583ec9ea8872ec7ef22c2452b43962daf13610c

Build ID: 8

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

dgl-bot · 2024-02-19T19:08:09Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

Co-authored-by: Andrzej Kotłowski <Andrzej.Kotlowski@intel.com>

dgl-bot · 2024-02-19T19:08:21Z

Commit ID: 2e53ac68ca21e27e7d9c112a26dc64fbbc09c726

Build ID: 35

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

dgl-bot · 2024-02-19T19:08:39Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-19T19:08:50Z

Commit ID: d3dae4c991ecf9b982b863b3bcf6ddad3e06b0c3

Build ID: 36

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

Co-authored-by: Andrzej Kotłowski <Andrzej.Kotlowski@intel.com>

dgl-bot · 2024-02-19T19:09:18Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-19T19:09:30Z

Commit ID: ab4331fcc8b18c3078c2c442be7a306eee1def8a

Build ID: 37

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

anko-intel · 2024-02-23T20:31:17Z

I have run ogb_example.py (base) and ogb_example_ARGO.py and observe following performance improvements:

Well done!
(measured on Ubuntu 22.04.4, DGL 2.0.0, pytorch 2.2.1)

Rhett-Ying · 2024-02-26T08:13:09Z

@dgl-bot

dgl-bot · 2024-02-26T08:58:29Z

Commit ID: a98604144e1065fff0c3b4c5237ad03e397e0418

Build ID: 38

Status: ✅ CI test succeeded.

Report path: link

Full logs path: link

frozenbugs · 2024-02-26T09:04:40Z

Can you separate out the tutorial part to another PR? For the example, we can merge directly as long as it is runnable with concrete readme.

dgl-bot · 2024-02-26T22:26:31Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-26T22:26:43Z

Commit ID: ee87efedd628c229e0aa0a79f27963dd408822b6

Build ID: 39

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

jasonlin316 · 2024-02-26T22:34:32Z

Can you separate out the tutorial part to another PR? For the example, we can merge directly as long as it is runnable with concrete readme.

No problem. I have removed the tutorial from this pull request, and created another one here: #7155

frozenbugs · 2024-02-27T09:43:48Z

It would be more impactful if you can add an example with graphbolt, our new dataloading package. graphbolt example: https://github.com/dmlc/dgl/blob/master/examples/sampling/graphbolt/node_classification.py

frozenbugs · 2024-02-27T09:43:59Z

@dgl-bot

examples/pytorch/argo/argo.py

dgl-bot · 2024-02-27T09:45:35Z

Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:

@dgl-bot

dgl-bot · 2024-02-27T09:45:46Z

Commit ID: 3ad4c2ea51b0820463ff5a1d9cdafb6eebed2e13

Build ID: 41

Status: ❌ CI test failed in Stage [Authentication].

Report path: link

Full logs path: link

frozenbugs · 2024-02-27T09:45:58Z

@dgl-bot

dgl-bot · 2024-02-27T09:47:02Z

Commit ID: d73aa7fb571ef3ebd05e837dbfc881f40f3aac63

Build ID: 40

Status: ⚪️ CI test cancelled due to overrun.

Report path: link

Full logs path: link

jasonlin316 added 5 commits January 23, 2024 12:01

add argo codes

d99bcb7

Add files via upload

81dd137

Update README.md

fbbea87

Update README.md

1172553

Update README.md

8bfa024

anko-intel reviewed Feb 1, 2024

View reviewed changes

DominikaJedynak reviewed Feb 1, 2024

View reviewed changes

examples/README.md Outdated Show resolved Hide resolved

Update README.md

ffa922d

Update README.md

cd7eb5a

jasonlin316 added 2 commits February 5, 2024 16:09

tmp remove to rename folder

de26574

rename folder to lower case

aa5f0e6

Update README.md

95b9cc7

Add files via upload

45bd1c1

Update README.md

186f01b

measure all epoch time

f92b035

Co-authored-by: Andrzej Kotłowski <Andrzej.Kotlowski@intel.com>

measure all epochs

048e1de

Co-authored-by: Andrzej Kotłowski <Andrzej.Kotlowski@intel.com>

jasonlin316 marked this pull request as ready for review February 19, 2024 19:09

anko-intel approved these changes Feb 19, 2024

View reviewed changes

frozenbugs self-requested a review February 26, 2024 09:03

DominikaJedynak approved these changes Feb 26, 2024

View reviewed changes

format code and remove the cpu tutorial

5d7a98d

frozenbugs approved these changes Feb 27, 2024

View reviewed changes

frozenbugs reviewed Feb 27, 2024

View reviewed changes

examples/pytorch/argo/argo.py Outdated Show resolved Hide resolved

Update examples/pytorch/argo/argo.py

4aaeefe

frozenbugs merged commit 2d2ad71 into dmlc:master Feb 27, 2024
1 of 2 checks passed

This pull request was closed.

[Feature] ARGO: an easy-to-use runtime to improve GNN training performance on multi-core processors #7003

[Feature] ARGO: an easy-to-use runtime to improve GNN training performance on multi-core processors #7003

Conversation

jasonlin316 commented Jan 23, 2024 • edited Loading

Description

Checklist

Changes

dgl-bot commented Jan 23, 2024

dgl-bot commented Jan 23, 2024

anko-intel left a comment

Choose a reason for hiding this comment

dgl-bot commented Feb 5, 2024

dgl-bot commented Feb 5, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 6, 2024

dgl-bot commented Feb 19, 2024

dgl-bot commented Feb 19, 2024

dgl-bot commented Feb 19, 2024

dgl-bot commented Feb 19, 2024

dgl-bot commented Feb 19, 2024

dgl-bot commented Feb 19, 2024

anko-intel commented Feb 23, 2024

Rhett-Ying commented Feb 26, 2024

dgl-bot commented Feb 26, 2024

frozenbugs commented Feb 26, 2024

dgl-bot commented Feb 26, 2024

dgl-bot commented Feb 26, 2024

jasonlin316 commented Feb 26, 2024

frozenbugs commented Feb 27, 2024

frozenbugs commented Feb 27, 2024

dgl-bot commented Feb 27, 2024

dgl-bot commented Feb 27, 2024

frozenbugs commented Feb 27, 2024

dgl-bot commented Feb 27, 2024

jasonlin316 commented Jan 23, 2024 •

edited

Loading