-
Notifications
You must be signed in to change notification settings - Fork 6
Cartesian Biweekly 2020 11 10
[ ] @Johann D Will make a GDP for looping
[x] @Johann D Will update Regions GDP with new implementation
[ ] @Oli F @Rhea G will propose changes to the quickstart based on fv3 and student experience
- Update 10.11.2020: Unfortunately no progress here. Currently all of our attention is on preparing the 3-day DSL workshop next week.
Review of action points of last meeting:
[x] Vulcan - does only removing sync in the stencil wrapper solve the problem?
Some tests passed, others did not (where numpy and cupy code linked stencils). Is a viable approach.
[x] @Eddie D Will request new reviews for PRs reviewed by Linus (who is away ATM)
[x] @Johann D Will make region temporary sizing change to regions implementation
Workshop been the Vulcan focus, so not much has happened in the last week, nor will happen in the next week.
Requires first removing or adding syncs for numpy and cupy code between stencil calls.
Expanded parameters - @Eddie D will write a GDP for this
def stencil(field_in: Field, fields: Tuple[Field, ...], *tracers: Field):
pass
gtscript.stencil(stencil, )
Local iterations - @Johann D will write a GDP for this (only compile-time int range for now)
def stencil(field_in: Field, fields: Tuple[Field, ...], *tracers: Field):
from __externals__ import nord
with computation(PARALLEL), interval(...):
for n in range(nord):
# do something in a loop
pass
Putting in the dycore. PR is here: https://github.com/GridTools/gt4py/pull/203
GDP is finalized. Implementation will finish up in the next few weeks.
Walkthrough series of PRs.
Demo.
Need to update GDP.
Kick-off meeting on Monday.
Goal: GTIR to code-generation for numpy and GT C++ backends (no CUDA)
Notebook is here:
This wiki is automatically updated from https://github.com/GridTools/concepts. See README on how to suggest changes.