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

hdf5 Bug Avoidance #75

Closed
wants to merge 23 commits into from
Closed

hdf5 Bug Avoidance #75

wants to merge 23 commits into from

Conversation

CodeGat
Copy link
Contributor

@CodeGat CodeGat commented Jul 3, 2024

This PR principally updates the spack-config used to deploy the model, so we avoid the buggy hdf5 1.13.3 that caused 2023.03.1 to be unreleased. The spack-packages repo was also updated to keep it relatively up to date, although the changes shouldn't have much of a bearing on the deployment.

References #73
Closes #74

Copy link
Contributor

github-actions bot commented Jul 3, 2024

🚀 Deploying access-om2 2024.03.2 as prerelease pr75-3

Details and usage instructions

This access-om2 model will be deployed as:

  • 2024.03.2 as a Release (when merged).
  • pr75-3 as a Prerelease (during this PR).

This Prerelease is accessible on Gadi using:

module use /g/data/vk83/prerelease/modules/access-models/
module load access-om2/pr75-3

where the binaries shall be on your $PATH.
This Prerelease is also accessible on Gadi via /g/data/vk83/prerelease/apps/spack/0.21/spack in the access-om2-pr75-3 environment.

🛠️ Using: spack-packages `2024.07.03`, spack-config `2024.07.02`
Details

It will be deployed using:

If this is not what was expected, commit changes to config/versions.json.

@CodeGat
Copy link
Contributor Author

CodeGat commented Jul 3, 2024

This current build is using hdf5 1.14.2 - https://github.com/ACCESS-NRI/ACCESS-OM2/actions/runs/9774408940/job/26982640737?pr=75#step:8:366

@dougiesquire
Copy link
Contributor

Thanks @CodeGat!

Some very basic testing indicates that the prerelease runs successfully but there is an impact on performance from these updates 😞

In the table below:

  • Exp 1 uses release-1deg_jra55_ryf modified to run for 1 year at a time;
  • Exp 2 uses the same configuration but the executable paths are changed to those built in prerelease pr75-3.

I ran three years of each. The experiments were run at the same time on Gadi. Times in the table are from the PAYU_WALLTIME field in job.yaml.

  Exp 1 Exp 2
output000 820.90 s 841.14 s
output001 830.14 s 850.29 s
output002 819.19 s 868.89 s
average 823.41 s 853.44

Below is the (selective) output of a diff between the two experiments, confirming that (this time) the executables were indeed changed in Exp 2 to the prerelease versions:

$ git diff -- om2_1deg_jra55_ryf_pr75-orig om2_1deg_jra55_ryf_pr75-3
diff --git a/om2_1deg_jra55_ryf_pr75-orig/config.yaml b/om2_1deg_jra55_ryf_pr75-3/config.yaml
index de774f2..3c38ae4 100644
--- a/om2_1deg_jra55_ryf_pr75-orig/config.yaml
+++ b/om2_1deg_jra55_ryf_pr75-3/config.yaml
@@ -30,7 +30,7 @@ input:
 submodels:
     - name: atmosphere
       model: yatm
-      exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/libaccessom2-git.2023.10.26=2023.10.26-ltfg7jcn6t4cefotvj3kjnyu5nru26xo/bin/yatm.exe
+      exe: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/libaccessom2-git.2023.10.26_2023.10.26-efhn2rufqy3orlpahmykegw5wxuowx3c/bin/yatm.exe
       input:
             - /g/data/vk83/experiments/inputs/access-om2/remapping_weights/JRA55/global.1deg/2020.05.30/rmp_jrar_to_cict_CONSERV.nc
             - /g/data/vk83/experiments/inputs/JRA-55/RYF/v1-4/data
@@ -38,7 +38,7 @@ submodels:

     - name: ocean
       model: mom
-      exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/mom5-git.2023.11.09=2023.11.09-qji4nlmr6utrribaiyhewe4je6mifguz/bin/fms_ACCESS-OM.x
+      exe: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/mom5-git.2023.11.09_2023.11.09-npjowklrqzkoz7xosbhuifyazjyrt5lk/bin/fms_ACCESS-OM.x
       input:
           - /g/data/vk83/experiments/inputs/access-om2/ocean/grids/mosaic/global.1deg/2020.05.30/grid_spec.nc
           - /g/data/vk83/experiments/inputs/access-om2/ocean/grids/mosaic/global.1deg/2020.05.30/ocean_hgrid.nc
@@ -57,7 +57,7 @@ submodels:

     - name: ice
       model: cice5
-      exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/cice5-git.2023.10.19=2023.10.19-v3zncpqjj2gyseudbwiudolcjq3k3leo/bin/cice_auscom_360x300_24x1_24p.exe
+      exe: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/cice5-git.2023.10.19_2023.10.19-itelnfmk22iwr6qszxjmgr4zwk7vmw3z/bin/cice_auscom_360x300_24x1_24p.exe
       input:
         - /g/data/vk83/experiments/inputs/access-om2/ice/grids/global.1deg/2020.05.30/grid.nc
         - /g/data/vk83/experiments/inputs/access-om2/ice/grids/global.1deg/2020.10.22/kmt.nc
@@ -74,7 +74,7 @@ collate:
   mem: 30GB
   ncpus: 4
   queue: normal
-  exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/mom5-git.2023.11.09=2023.11.09-qji4nlmr6utrribaiyhewe4je6mifguz/bin/mppnccombine.spack
+  exe: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/mom5-git.2023.11.09_2023.11.09-npjowklrqzkoz7xosbhuifyazjyrt5lk/bin/mppnccombine.spack

 # Misc
 runlog: false
diff --git a/om2_1deg_jra55_ryf_pr75-orig/manifests/exe.yaml b/om2_1deg_jra55_ryf_pr75-3/manifests/exe.yaml
index 80ed3f5..0d9571d 100644
--- a/om2_1deg_jra55_ryf_pr75-orig/manifests/exe.yaml
+++ b/om2_1deg_jra55_ryf_pr75-3/manifests/exe.yaml
@@ -2,17 +2,17 @@ format: yamanifest
 version: 1.0
 ---
 work/atmosphere/yatm.exe:
-  fullpath: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/libaccessom2-git.2023.10.26=2023.10.26-ltfg7jcn6t4cefotvj3kjnyu5nru26xo/bin/yatm.exe
+  fullpath: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/libaccessom2-git.2023.10.26_2023.10.26-efhn2rufqy3orlpahmykegw5wxuowx3c/bin/yatm.exe
   hashes:
-    binhash: 4e8b4ef76e971c4af3b26cfac632e160
-    md5: 5baa1d417fe6708fc30cbeaa57d82f96
+    binhash: b21a1ef267acae687c3c4cbc0ae51a8e
+    md5: f434bd5b517481ae27b4c8f47f95b94d
 work/ice/cice_auscom_360x300_24x1_24p.exe:
-  fullpath: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/cice5-git.2023.10.19=2023.10.19-v3zncpqjj2gyseudbwiudolcjq3k3leo/bin/cice_auscom_360x300_24x1_24p.exe
+  fullpath: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/cice5-git.2023.10.19_2023.10.19-itelnfmk22iwr6qszxjmgr4zwk7vmw3z/bin/cice_auscom_360x300_24x1_24p.exe
   hashes:
-    binhash: 3a65f67d21152e77034da28c22a94c66
-    md5: 37866455b057c85c3ea50c0ef0ea840b
+    binhash: 6f048474c92f2b2cc4ce14d159d9c3ad
+    md5: f74eb0ac0ca781c62d45c3a5e3f2fde7
 work/ocean/fms_ACCESS-OM.x:
-  fullpath: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/mom5-git.2023.11.09=2023.11.09-qji4nlmr6utrribaiyhewe4je6mifguz/bin/fms_ACCESS-OM.x
+  fullpath: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/mom5-git.2023.11.09_2023.11.09-npjowklrqzkoz7xosbhuifyazjyrt5lk/bin/fms_ACCESS-OM.x
   hashes:
-    binhash: 92ce1ff1a38f44f92ceafd67e8e7142c
-    md5: a3f10baeadb88e813b4a8121f61a6226
+    binhash: f4eaa918e5e24afa8050d4a5c9ae2382
+    md5: df8bec1f690c24cdfbeced2481352c38

@aidanheerdegen
Copy link
Member

This is roughly a 4% degradation in performance. My supposition is that it's the parallelio, as this showed a degradation in performance.

Options at this point:

  1. Live with the change
  2. Get more data (I'm not too concerned about 1 degree, but if it has a big impact on 0.1 that could be expensive)
  3. Try reverting parallelio and see if that fixes it
  4. Don't bother with this right now, and in the BGC case just add the bug fixes

@dougiesquire
Copy link
Contributor

dougiesquire commented Jul 4, 2024

I also tested a 0.1 deg configuration. In the table below:

  • Exp 1 uses release-01deg_jra55_ryf modified to run for 1 month at a time;
  • Exp 2 uses the same configuration but the executable paths are changed to those built in prerelease pr75-3.

I ran three months of each. The experiments were run at the same time on Gadi. Times in the table are from the PAYU_WALLTIME field in job.yaml.

  Exp 1 Exp 2
output000 4946.68 s 5040.16 s
output001 4517.40 s 4630.98 s
output002 4991.54 s 5098.09 s
average 4818.54 s 4923.07 s

So Exp 2 looks ~2-3% slower...

Below is the (selective) output of a diff between the two experiments:

$ git diff -- om2_01deg_jra55_ryf_pr75-orig om2_01deg_jra55_ryf_pr75-3
diff --git a/om2_01deg_jra55_ryf_pr75-orig/config.yaml b/om2_01deg_jra55_ryf_pr75-3/config.yaml
index f8c3249..6da9ffd 100644
--- a/om2_01deg_jra55_ryf_pr75-orig/config.yaml
+++ b/om2_01deg_jra55_ryf_pr75-3/config.yaml
@@ -28,7 +28,7 @@ input:
 submodels:
     - name: atmosphere
       model: yatm
-      exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/libaccessom2-git.2023.10.26=2023.10.26-ltfg7jcn6t4cefotvj3kjnyu5nru26xo/bin/yatm.exe
+      exe: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/libaccessom2-git.2023.10.26_2023.10.26-efhn2rufqy3orlpahmykegw5wxuowx3c/bin/yatm.exe
       input:
             - /g/data/vk83/experiments/inputs/access-om2/remapping_weights/JRA55/global.01deg/2020.05.30/rmp_jrar_to_cict_CONSERV.nc
             - /g/data/vk83/experiments/inputs/JRA-55/RYF/v1-4/data
@@ -36,7 +36,7 @@ submodels:

     - name: ocean
       model: mom
-      exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/mom5-git.2023.11.09=2023.11.09-qji4nlmr6utrribaiyhewe4je6mifguz/bin/fms_ACCESS-OM.x
+      exe: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/mom5-git.2023.11.09_2023.11.09-npjowklrqzkoz7xosbhuifyazjyrt5lk/bin/fms_ACCESS-OM.x
       input:
           - /g/data/vk83/experiments/inputs/access-om2/ocean/grids/mosaic/global.01deg/2020.05.30/grid_spec.nc
           - /g/data/vk83/experiments/inputs/access-om2/ocean/grids/mosaic/global.01deg/2020.05.30/ocean_hgrid.nc
@@ -55,7 +55,7 @@ submodels:

     - name: ice
       model: cice5
-      exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/cice5-git.2023.10.19=2023.10.19-v3zncpqjj2gyseudbwiudolcjq3k3leo/bin/cice_auscom_3600x2700_90x90_722p.exe
+      exe: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/cice5-git.2023.10.19_2023.10.19-itelnfmk22iwr6qszxjmgr4zwk7vmw3z/bin/cice_auscom_3600x2700_90x90_722p.exe
       input:
         - /g/data/vk83/experiments/inputs/access-om2/ice/grids/global.01deg/2020.05.30/grid.nc
         - /g/data/vk83/experiments/inputs/access-om2/ice/grids/global.01deg/2020.05.30/kmt.nc
diff --git a/om2_01deg_jra55_ryf_pr75-orig/manifests/exe.yaml b/om2_01deg_jra55_ryf_pr75-3/manifests/exe.yaml
index fe1c2e5..3e3de46 100644
--- a/om2_01deg_jra55_ryf_pr75-orig/manifests/exe.yaml
+++ b/om2_01deg_jra55_ryf_pr75-3/manifests/exe.yaml
@@ -2,17 +2,17 @@ format: yamanifest
 version: 1.0
 ---
 work/atmosphere/yatm.exe:
-  fullpath: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/libaccessom2-git.2023.10.26=2023.10.26-ltfg7jcn6t4cefotvj3kjnyu5nru26xo/bin/yatm.exe
+  fullpath: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/libaccessom2-git.2023.10.26_2023.10.26-efhn2rufqy3orlpahmykegw5wxuowx3c/bin/yatm.exe
   hashes:
-    binhash: 4e8b4ef76e971c4af3b26cfac632e160
-    md5: 5baa1d417fe6708fc30cbeaa57d82f96
+    binhash: b21a1ef267acae687c3c4cbc0ae51a8e
+    md5: f434bd5b517481ae27b4c8f47f95b94d
 work/ice/cice_auscom_3600x2700_90x90_722p.exe:
-  fullpath: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/cice5-git.2023.10.19=2023.10.19-v3zncpqjj2gyseudbwiudolcjq3k3leo/bin/cice_auscom_3600x2700_90x90_722p.exe
+  fullpath: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/cice5-git.2023.10.19_2023.10.19-itelnfmk22iwr6qszxjmgr4zwk7vmw3z/bin/cice_auscom_3600x2700_90x90_722p.exe
   hashes:
-    binhash: 67dd37413d389fc240c8ca24e3337fba
-    md5: f2787a34fcedaedc63a979c260e1c82d
+    binhash: 81832be2447772591f06152856921a42
+    md5: 7edfdd3d4d3c28109bb31321976389a8
 work/ocean/fms_ACCESS-OM.x:
-  fullpath: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/mom5-git.2023.11.09=2023.11.09-qji4nlmr6utrribaiyhewe4je6mifguz/bin/fms_ACCESS-OM.x
+  fullpath: /g/data/vk83/prerelease/apps/spack/0.21/release/linux-rocky8-x86_64/intel-2021.10.0/mom5-git.2023.11.09_2023.11.09-npjowklrqzkoz7xosbhuifyazjyrt5lk/bin/fms_ACCESS-OM.x
   hashes:
-    binhash: 92ce1ff1a38f44f92ceafd67e8e7142c
-    md5: a3f10baeadb88e813b4a8121f61a6226
+    binhash: f4eaa918e5e24afa8050d4a5c9ae2382
+    md5: df8bec1f690c24cdfbeced2481352c38

Copy link
Contributor

github-actions bot commented Jul 4, 2024

🚀 Deploying access-om2 2024.03.2 as prerelease pr75-4

Details and usage instructions

This access-om2 model will be deployed as:

  • 2024.03.2 as a Release (when merged).
  • pr75-4 as a Prerelease (during this PR).

This Prerelease is accessible on Gadi using:

module use /g/data/vk83/prerelease/modules/access-models/
module load access-om2/pr75-4

where the binaries shall be on your $PATH.
This Prerelease is also accessible on Gadi via /g/data/vk83/prerelease/apps/spack/0.21/spack in the access-om2-pr75-4 environment.

🛠️ Using: spack-packages `2024.07.03`, spack-config `2024.07.02`
Details

It will be deployed using:

If this is not what was expected, commit changes to config/versions.json.

@aidanheerdegen
Copy link
Member

So better (Exp 2 < 1% slower).

@aidanheerdegen are we okay with this?

I think it's your call. Well mostly your call. Yeah I think that is in the noise. Sorry I didn't see this earlier, we'd kinda decided to shelve the updates to the dependencies until we could work out what the issue was. Looks like you found it.

As long as that version of parallelio is fit for purpose then yeah, I think we should plough on and update everything but parallelio. It would be good to report something to the devs about impaired performance, but that would require a lot more work and who has the time?

I recall @anton-seaice had some issue with a parallelio bug in the MOM6-CICE6 world, but I suppose it doesn't affect OM2?

@anton-seaice
Copy link
Contributor

PIO 2.6.2 does a bit more validation of compatibility between dependencies and file formats than 2.5.2 did, this might cause some delay.

I haven't found any issues with 2.6.2, it did fix some issues we had with backward compatibility in 2.5.10 (NCAR/ParallelIO#1985).

@micaeljtoliveira found a floating point exception with hdf5, which we think has been fixed in the hdf5 main branch but not released yet. It would stop compilation when using debug flags, but normal release build compiles ok.

…s) and spack-config 2024.07.05 (remove om2 includes from base modules.yaml
Copy link
Contributor

github-actions bot commented Jul 8, 2024

🚀 Deploying access-om2 2024.03.2 as prerelease pr75-5

Details and usage instructions

This access-om2 model will be deployed as:

  • 2024.03.2 as a Release (when merged).
  • pr75-5 as a Prerelease (during this PR).

This Prerelease is accessible on Gadi using:

module use /g/data/vk83/prerelease/modules/access-models/
module load access-om2/pr75-5

where the binaries shall be on your $PATH.
This Prerelease is also accessible on Gadi via /g/data/vk83/prerelease/apps/spack/0.21/spack in the access-om2-pr75-5 environment.

🛠️ Using: spack-packages `2024.07.07`, spack-config `2024.07.05`
Details

It will be deployed using:

If this is not what was expected, commit changes to config/versions.json.

Copy link
Contributor

github-actions bot commented Jul 8, 2024

🚀 Deploying access-om2 2024.03.2 as prerelease pr75-6

Details and usage instructions

This access-om2 model will be deployed as:

  • 2024.03.2 as a Release (when merged).
  • pr75-6 as a Prerelease (during this PR).

This Prerelease is accessible on Gadi using:

module use /g/data/vk83/prerelease/modules/access-models/
module load access-om2/pr75-6

where the binaries shall be on your $PATH.
This Prerelease is also accessible on Gadi via /g/data/vk83/prerelease/apps/spack/0.21/spack in the access-om2-pr75-6 environment.

🛠️ Using: spack-packages `2024.07.07`, spack-config `2024.07.05`
Details

It will be deployed using:

If this is not what was expected, commit changes to config/versions.json.

Copy link
Contributor

github-actions bot commented Jul 8, 2024

🚀 Deploying access-om2 2024.03.2 as prerelease pr75-7

Details and usage instructions

This access-om2 model will be deployed as:

  • 2024.03.2 as a Release (when merged).
  • pr75-7 as a Prerelease (during this PR).

This Prerelease is accessible on Gadi using:

module use /g/data/vk83/prerelease/modules/access-models/
module load access-om2/pr75-7

where the binaries shall be on your $PATH.
This Prerelease is also accessible on Gadi via /g/data/vk83/prerelease/apps/spack/0.21/spack in the access-om2-pr75-7 environment.

🛠️ Using: spack-packages `2024.07.07`, spack-config `2024.07.05`
Details

It will be deployed using:

If this is not what was expected, commit changes to config/versions.json.

Copy link
Contributor

github-actions bot commented Jul 8, 2024

🚀 Deploying access-om2 2024.03.2 as prerelease pr75-7

Details and usage instructions

This access-om2 model will be deployed as:

  • 2024.03.2 as a Release (when merged).
  • pr75-7 as a Prerelease (during this PR).

This Prerelease is accessible on Gadi using:

module use /g/data/vk83/prerelease/modules/access-models/
module load access-om2/pr75-7

where the binaries shall be on your $PATH.
This Prerelease is also accessible on Gadi via /g/data/vk83/prerelease/apps/spack/0.21/spack in the access-om2-pr75-7 environment.

🛠️ Using: spack-packages `2024.07.07`, spack-config `2024.07.05`
Details

It will be deployed using:

If this is not what was expected, commit changes to config/versions.json.

Copy link
Contributor

github-actions bot commented Jul 8, 2024

🚀 Deploying access-om2 2024.03.2 as prerelease pr75-8

Details and usage instructions

This access-om2 model will be deployed as:

  • 2024.03.2 as a Release (when merged).
  • pr75-8 as a Prerelease (during this PR).

This Prerelease is accessible on Gadi using:

module use /g/data/vk83/prerelease/modules/access-models/
module load access-om2/pr75-8

where the binaries shall be on your $PATH.
This Prerelease is also accessible on Gadi via /g/data/vk83/prerelease/apps/spack/0.21/spack in the access-om2-pr75-8 environment.

🛠️ Using: spack-packages `2024.07.07`, spack-config `2024.07.05`
Details

It will be deployed using:

If this is not what was expected, commit changes to config/versions.json.

Copy link
Contributor

github-actions bot commented Jul 8, 2024

🚀 Deploying access-om2 2024.03.2 as prerelease pr75-9

Details and usage instructions

This access-om2 model will be deployed as:

  • 2024.03.2 as a Release (when merged).
  • pr75-9 as a Prerelease (during this PR).

This Prerelease is accessible on Gadi using:

module use /g/data/vk83/prerelease/modules/access-models/
module load access-om2/pr75-9

where the binaries shall be on your $PATH.
This Prerelease is also accessible on Gadi via /g/data/vk83/prerelease/apps/spack/0.21/spack in the access-om2-pr75-9 environment.

🛠️ Using: spack-packages `2024.07.08`, spack-config `2024.07.05`
Details

It will be deployed using:

If this is not what was expected, commit changes to config/versions.json.

@dougiesquire
Copy link
Contributor

In the table below:

  • Exp 1 uses release-1deg_jra55_ryf modified to run for 1 year at a time;
  • Exp 2 uses the same configuration but the executable paths are changed to those built in prerelease pr75-9

I ran three years of each. The experiments were run at the same time on Gadi. Times in the table are from the PAYU_WALLTIME field in job.yaml.

  Exp 1 Exp 2
output000 818.15 s 835.54 s
output001 827.26 s 833.76 s
output002 826.75 s 833.64 s
average 824.05 s 834.31 s

So the prerelease runs (~ 1% slower).

Below is the (selective) output of a diff between the two experiments:

$ git diff -- orig pr75-9
diff --git a/orig/config.yaml b/pr75-9/config.yaml
index de774f2..8e48816 100644
--- a/orig/config.yaml
+++ b/pr75-9/config.yaml
@@ -30,7 +30,7 @@ input:
 submodels:
     - name: atmosphere
       model: yatm
-      exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/libaccessom2-git.2023.10.26=2023.10.26-ltfg7jcn6t4cefotvj3kjnyu5nru26xo/bin/yatm.exe
+      exe: /g/data/vk83/prerelease/apps/spack/0.22/release/linux-rocky8-x86_64/intel-2021.10.0/libaccessom2-git.2023.10.26_2023.10.26-qotmoon5sfw6zp2hnjpntyjjpel7qi45/bin/yatm.exe
       input:
             - /g/data/vk83/experiments/inputs/access-om2/remapping_weights/JRA55/global.1deg/2020.05.30/rmp_jrar_to_cict_CONSERV.nc
             - /g/data/vk83/experiments/inputs/JRA-55/RYF/v1-4/data
@@ -38,7 +38,7 @@ submodels:

     - name: ocean
       model: mom
-      exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/mom5-git.2023.11.09=2023.11.09-qji4nlmr6utrribaiyhewe4je6mifguz/bin/fms_ACCESS-OM.x
+      exe: /g/data/vk83/prerelease/apps/spack/0.22/release/linux-rocky8-x86_64/intel-2021.10.0/mom5-git.2023.11.09_2023.11.09-ysj247r6owjfwcralshyy4yqpges7n6o/bin/fms_ACCESS-OM.x
       input:
           - /g/data/vk83/experiments/inputs/access-om2/ocean/grids/mosaic/global.1deg/2020.05.30/grid_spec.nc
           - /g/data/vk83/experiments/inputs/access-om2/ocean/grids/mosaic/global.1deg/2020.05.30/ocean_hgrid.nc
@@ -57,7 +57,7 @@ submodels:

     - name: ice
       model: cice5
-      exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/cice5-git.2023.10.19=2023.10.19-v3zncpqjj2gyseudbwiudolcjq3k3leo/bin/cice_auscom_360x300_24x1_24p.exe
+      exe: /g/data/vk83/prerelease/apps/spack/0.22/release/linux-rocky8-x86_64/intel-2021.10.0/cice5-git.2023.10.19_2023.10.19-3khrmuxma7p4wifa5av2biukx4iomk2l/bin/cice_auscom_360x300_24x1_24p.exe
       input:
         - /g/data/vk83/experiments/inputs/access-om2/ice/grids/global.1deg/2020.05.30/grid.nc
         - /g/data/vk83/experiments/inputs/access-om2/ice/grids/global.1deg/2020.10.22/kmt.nc
@@ -74,7 +74,7 @@ collate:
   mem: 30GB
   ncpus: 4
   queue: normal
-  exe: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/mom5-git.2023.11.09=2023.11.09-qji4nlmr6utrribaiyhewe4je6mifguz/bin/mppnccombine.spack
+  exe: /g/data/vk83/prerelease/apps/spack/0.22/release/linux-rocky8-x86_64/intel-2021.10.0/mom5-git.2023.11.09_2023.11.09-ysj247r6owjfwcralshyy4yqpges7n6o/bin/mppnccombine.spack

 # Misc
 runlog: false
diff --git a/orig/manifests/exe.yaml b/pr75-9/manifests/exe.yaml
index 80ed3f5..241395f 100644
--- a/orig/manifests/exe.yaml
+++ b/pr75-9/manifests/exe.yaml
@@ -2,17 +2,17 @@ format: yamanifest
 version: 1.0
 ---
 work/atmosphere/yatm.exe:
-  fullpath: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/libaccessom2-git.2023.10.26=2023.10.26-ltfg7jcn6t4cefotvj3kjnyu5nru26xo/bin/yatm.exe
+  fullpath: /g/data/vk83/prerelease/apps/spack/0.22/release/linux-rocky8-x86_64/intel-2021.10.0/libaccessom2-git.2023.10.26_2023.10.26-qotmoon5sfw6zp2hnjpntyjjpel7qi45/bin/yatm.exe
   hashes:
-    binhash: 4e8b4ef76e971c4af3b26cfac632e160
-    md5: 5baa1d417fe6708fc30cbeaa57d82f96
+    binhash: d14806ad4b0b890d4c2f78a7e34258ab
+    md5: 651ea586afa441f3d94376f28f26b54d
 work/ice/cice_auscom_360x300_24x1_24p.exe:
-  fullpath: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/cice5-git.2023.10.19=2023.10.19-v3zncpqjj2gyseudbwiudolcjq3k3leo/bin/cice_auscom_360x300_24x1_24p.exe
+  fullpath: /g/data/vk83/prerelease/apps/spack/0.22/release/linux-rocky8-x86_64/intel-2021.10.0/cice5-git.2023.10.19_2023.10.19-3khrmuxma7p4wifa5av2biukx4iomk2l/bin/cice_auscom_360x300_24x1_24p.exe
   hashes:
-    binhash: 3a65f67d21152e77034da28c22a94c66
-    md5: 37866455b057c85c3ea50c0ef0ea840b
+    binhash: 000959a21e84ba4baa66c32e61c03d39
+    md5: 237c0c1838fd072e43f7b632315cc2bb
 work/ocean/fms_ACCESS-OM.x:
-  fullpath: /g/data/vk83/apps/spack/0.20/release/linux-rocky8-x86_64/intel-19.0.5.281/mom5-git.2023.11.09=2023.11.09-qji4nlmr6utrribaiyhewe4je6mifguz/bin/fms_ACCESS-OM.x
+  fullpath: /g/data/vk83/prerelease/apps/spack/0.22/release/linux-rocky8-x86_64/intel-2021.10.0/mom5-git.2023.11.09_2023.11.09-ysj247r6owjfwcralshyy4yqpges7n6o/bin/fms_ACCESS-OM.x
   hashes:
-    binhash: 92ce1ff1a38f44f92ceafd67e8e7142c
-    md5: a3f10baeadb88e813b4a8121f61a6226
+    binhash: 66f01cf1c5e589ba8a253bc09bb365c7
+    md5: 04e7b7594760b7072ff828dd463acf5e

dougiesquire
dougiesquire previously approved these changes Jul 9, 2024
@dougiesquire
Copy link
Contributor

We decided not to merge this yet, given that there still seems to be a performance hit and further investigation is needed to work out what is causing this.

@CodeGat
Copy link
Contributor Author

CodeGat commented Sep 24, 2024

Going to have to rebase this PR to the new main - done some history rewriting here.

@CodeGat CodeGat marked this pull request as draft October 23, 2024 21:39
@CodeGat
Copy link
Contributor Author

CodeGat commented Oct 30, 2024

Will be closing this for now. The PR and branch will be here for future work, and it can always be reopened.

@CodeGat CodeGat closed this Oct 30, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2024.03.1 release is broken
4 participants