idaholab
KRUSTY Multiphysics Model
Job VTB Documentation on fef95d6 wanted to post the following:
View the site here
This comment will be updated on new commits.
@GiudGiud Thanks for the review. I think we resolved all the comments. It should be ready for another round of review.
I added some syntax check tests and mesh gen tests. Not sure if we want regression tests as these are kind of heavy.
can you run a single timestep for the BISON case? and what about an eigenvalue calculation without multiphysics for the Griffin one? how expensive would each of those be? we have access to a beefier machine for Griffin now, within reason we can make some larger cases work
The Griffin neutronic alone calculation on sawtooth is roughly about 12 minutes with 640 CPUs if using NA=3, SN35. For NA=1 and SN13, the run time can be 20 times faster. Yan
From: Guillaume Giudicelli @.> Sent: Wednesday, July 10, 2024 4:18 PM To: idaholab/virtual_test_bed @.> Cc: Cao, Yan @.>; Assign @.> Subject: Re: [idaholab/virtual_test_bed] KRUSTY Multiphysics Model (PR #404)
can you run a single timestep for the BISON case? and what about an eigenvalue calculation without multiphysics for the Griffin one? how expensive would each of those be? we have access to a beefier machine for Griffin now, within reason we ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd
can you run a single timestep for the BISON case? and what about an eigenvalue calculation without multiphysics for the Griffin one? how expensive would each of those be? we have access to a beefier machine for Griffin now, within reason we can make some larger cases work
— Reply to this email directly, view it on GitHubhttps://urldefense.us/v3/__https:/github.com/idaholab/virtual_test_bed/pull/404*issuecomment-2221496270__;Iw!!G_uCfscf7eWS!dpTHmxfrOxit8_Zfl7FMbY_AcgW47jq_7pd5L-eDIk2eqAzok3xEPVX11Lq3AxSKcGAd_vr7ceYv7hsGrdL91w$, or unsubscribehttps://urldefense.us/v3/__https:/github.com/notifications/unsubscribe-auth/BJUPMMEVCCIEDORCLZSXJKTZLWQH3AVCNFSM6AAAAABKKBIQO6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMRRGQ4TMMRXGA__;!!G_uCfscf7eWS!dpTHmxfrOxit8_Zfl7FMbY_AcgW47jq_7pd5L-eDIk2eqAzok3xEPVX11Lq3AxSKcGAd_vr7ceYv7hsPQQen5w$. You are receiving this because you were assigned.Message ID: @.@.>>
For NA=1 and SN13, the run time can be 20 times faster.
that seems workable. The tests can modify the input through cli_args so we can afford to run that for regression
@GiudGiud just a friendly reminder that this PR is ready for another round of review. Thanks.
Hello!
I'll get to it once our giant submodule update PR makes it through. Thansk for reaching it out
@GiudGiud Just a friendly reminder that this is still pending review. Thanks in advance.
to reduce the picture size you must remove it from the commit that introduced it. Otherwise it does not change anything with regards to the repository size
to reduce the picture size you must remove it from the commit that introduced it. Otherwise it does not change anything with regards to the repository size
Will do. I am working on the other comments and will do a rebase once everything looks fine.
looks like it does not converge on HPC for me. We could be getting unlucky, what is the exact configuration you ran? 15 nodes * 40 procs?
alternatively we should change BISON to not error but rather to use the solutionInvalidInterface here
I used one sawtooth node (48 cores) for the short testing problem here.
For full problem, I use 10 nodes * 24 cores on sawtooth for SN(2,3) NA=3. I will double check the input...
for the input as is (not changing the azimuthal, so SN 3,5,3) what did you use?
I imagine 2,3,3 works fine?
@aabdelhameedd what was the computational resources configuration to run the input as it is currently ?
I think we only use 3 5 3 for neutronics only. How many nodes did you use? @ycao-anl
I used 640 CPUs.
Yan
From: Yinbin Miao @.> Sent: Monday, August 19, 2024 11:44 AM To: idaholab/virtual_test_bed @.> Cc: Cao, Yan @.>; Mention @.> Subject: Re: [idaholab/virtual_test_bed] KRUSTY Multiphysics Model (PR #404)
I think we only use 3 5 3 for neutronics only. How many nodes did you use? @ycao-anl — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned. Message ID: <idaholab/virtual_test_bed/pull/404/c2296997762@ github. com> ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd
I think we only use 3 5 3 for neutronics only. How many nodes did you use? @ycao-anlhttps://urldefense.us/v3/__https:/github.com/ycao-anl__;!!G_uCfscf7eWS!ZgqJOEp9MbrDFJy7F7-aa7SGsg63cQHj_tI4x_SqS511TZ8O44QVRCnbmYnRvVkoivz8luUi95s_R03k8_ldiw$
— Reply to this email directly, view it on GitHubhttps://urldefense.us/v3/__https:/github.com/idaholab/virtual_test_bed/pull/404*issuecomment-2296997762__;Iw!!G_uCfscf7eWS!ZgqJOEp9MbrDFJy7F7-aa7SGsg63cQHj_tI4x_SqS511TZ8O44QVRCnbmYnRvVkoivz8luUi95s_R03JxQxXvQ$, or unsubscribehttps://urldefense.us/v3/__https:/github.com/notifications/unsubscribe-auth/BJUPMMEZ65UQ2KJZHJXPWSLZSIOELAVCNFSM6AAAAABKKBIQO6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEOJWHE4TONZWGI__;!!G_uCfscf7eWS!ZgqJOEp9MbrDFJy7F7-aa7SGsg63cQHj_tI4x_SqS511TZ8O44QVRCnbmYnRvVkoivz8luUi95s_R02H0-IR2w$. You are receiving this because you were mentioned.Message ID: @.@.>>
how did you do 640 on sawtooth? 16 nodes with 40 cpus each? asking because sawtooth has 48 cpus per node
Ah, for the multiphysics simulations, the power needs to be reduced as the model is for the warm critical tests without heat pipe heat removal. I will fix it soon. We use 0.01 W for steady state as the power only rises to ~900W during reactivity insertion transient.
Yes, I used 40 cpus and 16 nodes.
From: Guillaume Giudicelli @.> Sent: Monday, August 19, 2024 11:58 AM To: idaholab/virtual_test_bed @.> Cc: Cao, Yan @.>; Mention @.> Subject: Re: [idaholab/virtual_test_bed] KRUSTY Multiphysics Model (PR #404)
how did you do 640 on sawtooth? 16 nodes with 40 cpus each? asking because sawtooth has 48 cpus per node — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned. Message ID: ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd
how did you do 640 on sawtooth? 16 nodes with 40 cpus each? asking because sawtooth has 48 cpus per node
— Reply to this email directly, view it on GitHubhttps://urldefense.us/v3/__https:/github.com/idaholab/virtual_test_bed/pull/404*issuecomment-2297022681__;Iw!!G_uCfscf7eWS!b5YoHGT_FJQnC-dORaYKOLBcLDlUsXN8yubFRK9C8w3z0RrBwoghFk4TtxZ4Em1DFrQBr0kiyB66E53tHPIxQw$, or unsubscribehttps://urldefense.us/v3/__https:/github.com/notifications/unsubscribe-auth/BJUPMMDKHUI6WHO7OQGD6NTZSIPZ5AVCNFSM6AAAAABKKBIQO6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEOJXGAZDENRYGE__;!!G_uCfscf7eWS!b5YoHGT_FJQnC-dORaYKOLBcLDlUsXN8yubFRK9C8w3z0RrBwoghFk4TtxZ4Em1DFrQBr0kiyB66E533XIpJvg$. You are receiving this because you were mentioned.Message ID: @.@.>>
Sounds good. Thanks @ycao-anl Thanks @miaoyinb we should be able to roll this out soon then! VTB testing is back on now too
I'll take it from here @miaoyinb I was waiting for the recipes to update. I think you merged it right on time though
Test failures are unrelated, on other modles, and being addressed in current maintenance operations