Add `dev-01deg_jra55_ryf+wombatlite` configuration by dougiesquire · Pull Request #286 · ACCESS-NRI/access-om2-configs

dougiesquire · 2026-03-07T10:07:46Z

This PR:

Adds changes to configure 0.1deg RYF WOMBATlite. Contributes to Add WOMBATlite configurations #270
Adds cost info to README. Contributes to Add cost of each config to its README #287
Removes .travis.yml file. Contributes to Remove .travis.yml from 0.1 deg configs #325

This is preparation for the first official release of this configuration.

Still to do:

Move inputs out of prerelease area
Add comment to redundant section of field_table
Update performance info
Update checksums

dougiesquire · 2026-03-07T10:09:45Z

!test repro commit

github-actions · 2026-03-07T10:42:09Z

❌ The Bitwise Reproducibility Check Failed ❌

When comparing:

270-01deg_jra55_ryf+wombatlite (checksums created using commit 21aad8a), against
dev-01deg_jra55_ryf+wombatlite (checksums in commit 6c759fd)

🔧 The new checksums will be committed to this PR, if they differ from what is on this branch.

Further information

The experiment can be found on Gadi at /scratch/tm70/repro-ci/experiments/access-om2-configs/pr286/270-01deg_jra55_ryf+wombatlite/21aad8a0a9cce9674588aec94550be8f3f8f3e11, and the test results at https://github.com/ACCESS-NRI/access-om2-configs/runs/66134313760.

The checksums generated by this !test command are found in the testing/checksum directory of https://github.com/ACCESS-NRI/access-om2-configs/actions/runs/22797073144/artifacts/5810794199.

The checksums compared against are found here https://github.com/ACCESS-NRI/access-om2-configs/tree/6c759fdcf40b08d256bcef88ec387ac5abad4ef4/testing/checksum

Test summary:
❌ test_repro_historical
✅ test_repro_determinism
✅ test_repro_restart

dougiesquire · 2026-03-10T22:39:02Z

This does not complete 1 month within the 5 hour walltime limit (the max for this number of CPUs).

Approximately:

WOMBAT legacy (10 tr, ~10 equations) is about 1.8-1.9x slower than running without BGC. Just adding 10 passive tracers slows the model by about 1.6x.
WOMBATlite (15 tr, ~100 equations) is about 3-3.4x slower than running without BGC. Just adding 15 passive tracers slows the model by about 2x.

For reference:

BLING (7 tr, ~50 equations) is about 2x slower than running without BGC. Just adding 7 passive tracers slows the model by about 1.4x.

There are some low-hanging opportunities for code refactoring to improve WOMBATlite performance, but so far I've only been able to squeeze out ~7%.

access-hive-bot · 2026-03-11T05:02:20Z

This pull request has been mentioned on ACCESS Hive Community Forum. There might be relevant details there:

https://forum.access-hive.org.au/t/cosima-twg-meeting-minutes-2026/5906/8

Relative to non-BGC config: Before this change: - 3.5x slower - 3.5x costlier After this change: - 2.0x slower - 2.6x costlier

dougiesquire · 2026-03-12T03:44:29Z

!test repro commit

dougiesquire · 2026-03-12T03:47:58Z

This does not complete 1 month within the 5 hour walltime limit (the max for this number of CPUs).

I've moved this configuration to Sapphire Rapids nodes and increased number of CPUs used by the ocean (see 6056897). With these changes this configuration is 2.0x slower and 2.6x costlier that the equivalent non-BGC config (previously 3.5x the speed and cost). 1 month now easily completes within the walltime limit.

github-actions · 2026-03-12T04:25:34Z

❌ !test Command Failed ❌ See https://github.com/ACCESS-NRI/access-om2-configs/actions/runs/22985609834

dougiesquire · 2026-03-12T05:02:43Z

!test repro commit

github-actions · 2026-03-12T05:33:21Z

❌ The Bitwise Reproducibility Check Failed ❌

When comparing:

270-01deg_jra55_ryf+wombatlite (checksums created using commit 8109a11), against
dev-01deg_jra55_ryf+wombatlite (checksums in commit 6c759fd)

🔧 The new checksums will be committed to this PR, if they differ from what is on this branch.

Further information

The experiment can be found on Gadi at /scratch/tm70/repro-ci/experiments/access-om2-configs/pr286/270-01deg_jra55_ryf+wombatlite/8109a11dc41b40dfb45c1f1eaf7bbd6ba71348e5, and the test results at https://github.com/ACCESS-NRI/access-om2-configs/runs/66742886479.

The checksums generated by this !test command are found in the testing/checksum directory of https://github.com/ACCESS-NRI/access-om2-configs/actions/runs/22987403441/artifacts/5884709069.

The checksums compared against are found here https://github.com/ACCESS-NRI/access-om2-configs/tree/6c759fdcf40b08d256bcef88ec387ac5abad4ef4/testing/checksum

Test summary:
❌ test_repro_historical
✅ test_repro_determinism
✅ test_repro_restart

…-configs/actions/runs/22987403441

dougiesquire · 2026-03-13T02:18:10Z

@anton-seaice this is ready for review. It probably makes sense to review #280 first as I've added some comments there that may help answer some questions that might come up.

This configuration uses 62 SR nodes, with no unused CPUs (down from 7 unused CL CPUs)

The failing QA checks will be fixed by ACCESS-NRI/model-config-tests#204

dougiesquire · 2026-03-16T04:50:23Z

!test repro commit

github-actions · 2026-03-16T05:21:34Z

❌ The Bitwise Reproducibility Check Failed ❌

When comparing:

270-01deg_jra55_ryf+wombatlite (checksums created using commit 6479ddf), against
dev-01deg_jra55_ryf+wombatlite (checksums in commit 6c759fd)

🔧 The new checksums will be committed to this PR, if they differ from what is on this branch.

Further information

The experiment can be found on Gadi at /scratch/tm70/repro-ci/experiments/access-om2-configs/pr286/270-01deg_jra55_ryf+wombatlite/6479ddfcf473a6c389f660a244682ab0a60998cc, and the test results at https://github.com/ACCESS-NRI/access-om2-configs/runs/67178662554.

The checksums generated by this !test command are found in the testing/checksum directory of https://github.com/ACCESS-NRI/access-om2-configs/actions/runs/23128443861/artifacts/5937557785.

The checksums compared against are found here https://github.com/ACCESS-NRI/access-om2-configs/tree/6c759fdcf40b08d256bcef88ec387ac5abad4ef4/testing/checksum

Test summary:
❌ test_repro_historical
✅ test_repro_determinism
✅ test_repro_restart

…-configs/actions/runs/23128443861

dougiesquire · 2026-06-03T23:27:15Z

!test repro commit

github-actions · 2026-06-04T00:02:23Z

❌ The Bitwise Reproducibility Check Failed ❌

When comparing:

270-01deg_jra55_ryf+wombatlite (checksums created using commit b4246c1), against
dev-01deg_jra55_ryf+wombatlite (checksums in commit 6c759fd)

🔧 The new checksums will be committed to this PR, if they differ from what is on this branch.

Further information

The experiment can be found on Gadi at /scratch/tm70/repro-ci/experiments/access-om2-configs/pr286/270-01deg_jra55_ryf+wombatlite/b4246c129e7d14449788b2d2b0c8b17fe3d3ac27, and the test results at https://github.com/ACCESS-NRI/access-om2-configs/runs/79421011290.

The checksums generated by this !test command are found in the testing/checksum directory of https://github.com/ACCESS-NRI/access-om2-configs/actions/runs/26919516182/artifacts/7399648563.

The checksums compared against are found here https://github.com/ACCESS-NRI/access-om2-configs/tree/6c759fdcf40b08d256bcef88ec387ac5abad4ef4/testing/checksum

Test summary:
❌ test_repro_historical
✅ test_repro_determinism
✅ test_repro_restart
✅ test_repro_payu_setup

…-configs/actions/runs/26919516182

dougiesquire · 2026-06-04T01:11:46Z

I've run this for 6 months and Pearse and I are happy with the WOMBAT output

Configure WOMBATlite

21aad8a

Run on Sapphire Rapids nodes

6056897

Relative to non-BGC config: Before this change: - 3.5x slower - 3.5x costlier After this change: - 2.0x slower - 2.6x costlier

dougiesquire force-pushed the 270-01deg_jra55_ryf+wombatlite branch from e2b853d to 6056897 Compare March 12, 2026 03:34

Add Performance section to README

8109a11

Add comment to redundant section of field_table

5b153bd

dougiesquire self-assigned this Mar 12, 2026

access-bot and others added 2 commits March 12, 2026 05:33

Updated checksums as part of https://github.com/ACCESS-NRI/access-om2…

a8f6a13

…-configs/actions/runs/22987403441

Move inputs out of prerelease area

2fd1b3e

dougiesquire marked this pull request as ready for review March 13, 2026 02:17

dougiesquire requested a review from anton-seaice March 13, 2026 02:18

Update to access-om2/2026.02.002

6479ddf

Updated checksums as part of https://github.com/ACCESS-NRI/access-om2…

26069bd

…-configs/actions/runs/23128443861

anton-seaice previously approved these changes Mar 16, 2026

View reviewed changes

dougiesquire dismissed anton-seaice’s stale review via 8b9dd21 June 3, 2026 03:10

dougiesquire force-pushed the 270-01deg_jra55_ryf+wombatlite branch from 8b9dd21 to 1c270f7 Compare June 3, 2026 03:31

Update to access-om2/2026.05.000

9fb1862

dougiesquire force-pushed the 270-01deg_jra55_ryf+wombatlite branch from 1c270f7 to 9fb1862 Compare June 3, 2026 04:31

Remove .travis.yml file

26d9638

dougiesquire added 2 commits June 3, 2026 15:14

Update runoff concentrations in data_table

ed597d4

Update performance info in README

b4246c1

Updated checksums as part of https://github.com/ACCESS-NRI/access-om2…

7448c76

…-configs/actions/runs/26919516182

dougiesquire mentioned this pull request Jun 4, 2026

Remove .travis.yml from 0.1 deg configs #325

Open

dougiesquire requested a review from anton-seaice June 4, 2026 01:11

Conversation

dougiesquire commented Mar 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dougiesquire commented Mar 7, 2026

Uh oh!

github-actions Bot commented Mar 7, 2026

Uh oh!

dougiesquire commented Mar 10, 2026

Uh oh!

access-hive-bot commented Mar 11, 2026

Uh oh!

dougiesquire commented Mar 12, 2026

Uh oh!

dougiesquire commented Mar 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Mar 12, 2026

Uh oh!

dougiesquire commented Mar 12, 2026

Uh oh!

github-actions Bot commented Mar 12, 2026

Uh oh!

dougiesquire commented Mar 13, 2026

Uh oh!

dougiesquire commented Mar 16, 2026

Uh oh!

github-actions Bot commented Mar 16, 2026

Uh oh!

dougiesquire commented Jun 3, 2026

Uh oh!

github-actions Bot commented Jun 4, 2026

Uh oh!

dougiesquire commented Jun 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

dougiesquire commented Mar 7, 2026 •

edited

Loading

dougiesquire commented Mar 12, 2026 •

edited

Loading