Run the hello-global-ocean example in Docs #542
Conversation
navidcy
added
documentation
|
yah! because we run on tartarus now! |
|
Seems to hit memory limit during Hm... |
|
Somehow it's trying to create the ETOPO native grid on the GPU! |
|
I'm wondering why we don't hit the same issue with the near-global example |
|
I guess it's a bit too much memory-wise? I'm wondering if some of the defaults trigger bigger memory allocation than the near-global case in the examples. I'm closing the PR. Feel free to re-open or we can try later? |
|
Probably the issue is that we are overloading the (smallish) Tartarus GPU with the examples? |
|
hm.... perhaps.... I'm trying to run the hello-global-ocean on tartarus at the moment (just in the REPL) to see if it goes through. |
|
btw, I run the README hello-global-ocean example on tartarus on
Looks really strange isn't it? |
|
@glwagner you sure you didn't use higher vertical resolution for the image we have in the README? The snapshot I got looks really bad compared with the one in the README... |
|
The difference is the initial conditions I think. In the README we were starting from ECCO2 which is eddying, then the default changed to ECCO4 that is not eddying, so no eddies in that small integration time. |
|
oh I can change that and try again! thanks! |
|
hm... before the Metadata refactor the README was initialised with But I'll try with ECCO2 and see if it's different. |
|
Nono, that change was much before the Metadata refactor |
|
OK great. I'll run with ECCO2! |
|
ok, reproduced this: |
|
docs with the example at https://clima.github.io/ClimaOceanDocumentation/previews/PR542/ |
No description provided.