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.

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

Fix/restart gpu backend #329

Merged
merged 12 commits into from
Sep 20, 2022
Merged

Conversation

elynnwu
Copy link
Collaborator

@elynnwu elynnwu commented Sep 16, 2022

This PR was used to run dace gpu orchestration on daint for the TC restart case for 10 days.

@review-notebook-app
Copy link

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.


Powered by ReviewNB

@elynnwu elynnwu marked this pull request as ready for review September 20, 2022 18:20
@elynnwu elynnwu changed the base branch from feature/fortran_restart to main September 20, 2022 18:27
@elynnwu elynnwu changed the base branch from main to feature/fortran_restart September 20, 2022 18:27
@ajdas1 ajdas1 merged commit 69fc91b into feature/fortran_restart Sep 20, 2022
@ajdas1 ajdas1 deleted the fix/restart_gpu_backend branch September 20, 2022 19:09
FlorianDeconinck added a commit that referenced this pull request Dec 14, 2022
* initial commit

* first attempt at functions to restart from Fortran

* added code in _restart_driver_state to call _overwrite_state_from_fortran

* fixed state traversal

* restartconfig now runs, but model crashes after

* lots of print statements

* removed print statements - mostly

* modifying properties

* small changes

* updated properties

* updated RESTART_PROPERTIES

* updated gt4py, buildenv

* renamed pace physics

* pre-commit

* writing test for restart

* check only dycore, tests

* modified properties

* begin generating grid from metric terms

* git pull on main

* add from_lat_lon method to MetricTerms

* metric_terms look good

* testing metric terms

* gitignore

* next step

* added grid transformation - Schmidt

* updated history

* lint

* testing ak, bk

* gitignore removed nc, png

* added grid config to driver

* pre-commit

* changed default to None

* added grid_config to test_driver.

* this doesn't work

* reads tc restart - no grid transform

* small changes

* stretched grid testing

* next steps

* moved offset factor to function

* consolidate driver functions

* config

* main gitignore

* setting up TropicalCycloneConfig

* added stuff

* working on init_tc_state

* tc_init mostly done

* pre-commit

* update

* update

* uncommented some stuff

* fixed wind init

* testing

* test

* somehow still nans after ts0

* trying to find where things go wrong

* empty

* second try

* Fix/restart gpu backend (#329)

* update default namelist option

* init works with dace gpu, but orch broken

* removed type hinting in stretch transform

Co-authored-by: Ajda Savarin <[email protected]>

* typing

* cleaned up

* pre-commit

* removed some notebooks

* Update .gitignore

Co-authored-by: Jeremy McGibbon <[email protected]>

* updated path to google bucket with restart files

* updated stretch_transform with np, removed sign

* exchanged numpy_module for np

* updated type hints on stretch_transform

* added np to doc string

* removed empty lines from _legacy_restart

* removed stretch_grid from fv3core config

* added fortran restart config

* restart_driver_state determines if fortran or not

* started working on GridInitializer for generated

* in driver, separate grid and restart in, beginning

* self.config.grid?

* get grid init separate from driver state init

* removed ptop from vertical grid data

* removing fortran_restart flag

* RESTART_PROPERTIES reverted, added to driver

* undid google cloud bucket path

* making legacy_restart work ...

* pre-commit

* another pre-commit

* changed namelist to not dycore only

* gitignore

* allows to download data from google bucket

* grid_init changed to grid_config

* grid config

* docstrings

* refactor grid functions to take in more arguments

* hardcode default stretch_factor, lon, lat target

* serialized_grid boolean removed

* more vertical grid data changes

* removed gridconfig

* docstring for update fortran restart pe_peln

* update for different np

* renoved return statement from dict_state_to_driver

* removed return for metric terms

* removed some return statements

* cool new syntax if file has string

* for more than 6 ranks, scatter state

* minor

* moved fortran restart from state to initialization

* tracer properties

* refactor baroclinic_init to take in grid data

* pre-commit

* moved fortran stuff back to state

* gitignore

* pace util tests for test_legacy_restart

* tests and baroclinic restart work

* pre-commit

* remove comments

* initialization

* removed comments

* deleted jupyter notebook

* pre-commit

* metric terms to grid data

* fixing the merge

* Fixing merge

* Clean up PR

* Cleanup PR

* Here be dragons warning

* Revert uneeded changes to state

* Revert more uneeeded change
Lint

* Revert uneeded change

* Revert assert

* Exclude tropical configs from test

Co-authored-by: oelbert <[email protected]>
Co-authored-by: Elynn Wu <[email protected]>
Co-authored-by: Jeremy McGibbon <[email protected]>
Co-authored-by: Florian Deconinck <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants