From 497019d8c8d647e7f10215e96c45734878699878 Mon Sep 17 00:00:00 2001 From: Caspar van Leeuwen Date: Tue, 5 Aug 2025 16:24:03 +0200 Subject: [PATCH 1/4] Rebuild all CUDA software with EB-5.1.1 now that we have a CUDA sanity check, so we can see if anything is 'broken'. Also, there are so many 'holes' in which software is present for which combination of CPU+GPU, that this is a convenient way to fill the gaps --- ...5-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml | 13 +++++++++++++ ...5-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml | 14 ++++++++++++++ 2 files changed, 27 insertions(+) create mode 100644 easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml create mode 100644 easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml diff --git a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml new file mode 100644 index 0000000000..ab1bef8e74 --- /dev/null +++ b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml @@ -0,0 +1,13 @@ +# We'll rebuild all CUDA software, for various reasons +# 1. We now have a proper CUDA sanity check, and if anything was 'wrong' with our current CUDA installs, we'd like +# to know about it +# 2. The PR implementing a CI to check for differences between officially supported CUDA Compute Capabilities shows +# that there are a lot of missing installations https://github.com/EESSI/software-layer/pull/1087 . A rebuild PR like +# this will have the convenient side effect of filling all those holes +easyconfigs: + - CUDA-12.1.1.eb + - cuDNN-8.9.2.26-CUDA-12.1.1.eb + - LAMMPS-2Aug2023_update2-foss-2023a-kokkos-CUDA-12.1.1.eb + - ESPResSo-4.2.2-foss-2023a-CUDA-12.1.1.eb + - LightGBM-4.5.0-foss-2023a-CUDA-12.1.1.eb + diff --git a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml new file mode 100644 index 0000000000..c9f1f9b47f --- /dev/null +++ b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml @@ -0,0 +1,14 @@ +# We'll rebuild all CUDA software, for various reasons +# 1. We now have a proper CUDA sanity check, and if anything was 'wrong' with our current CUDA installs, we'd like +# to know about it +# 2. The PR implementing a CI to check for differences between officially supported CUDA Compute Capabilities shows +# that there are a lot of missing installations https://github.com/EESSI/software-layer/pull/1087 . A rebuild PR like +# this will have the convenient side effect of filling all those holes +easyconfigs: + - CUDA-12.4.0.eb: + options: + accept-eula-for: CUDA + - UCX-CUDA-1.15.0-GCCcore-13.2.0-CUDA-12.4.0.eb + - UCC-CUDA-1.2.0-GCCcore-13.2.0-CUDA-12.4.0.eb + - OSU-Micro-Benchmarks-7.5-gompi-2023b-CUDA-12.4.0.eb + - GROMACS-2024.4-foss-2023b-CUDA-12.4.0.eb From b7f7b0a5edfb719cf2e925142398d5416e0459b0 Mon Sep 17 00:00:00 2001 From: Caspar van Leeuwen Date: Tue, 5 Aug 2025 16:49:58 +0200 Subject: [PATCH 2/4] Rename, now with 'eb' in the name --- ... => 20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml} | 0 ... => 20250805-eb-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml} | 0 2 files changed, 0 insertions(+), 0 deletions(-) rename easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/{20250805-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml => 20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml} (100%) rename easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/{20250805-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml => 20250805-eb-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml} (100%) diff --git a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml similarity index 100% rename from easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml rename to easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml diff --git a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml similarity index 100% rename from easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml rename to easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml From 36c88788bf399b20d3e7dd8b3b031c781ae7b194 Mon Sep 17 00:00:00 2001 From: Caspar van Leeuwen Date: Wed, 6 Aug 2025 11:19:29 +0200 Subject: [PATCH 3/4] Make sure to accept EULA --- .../20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml index ab1bef8e74..5d701e9eb9 100644 --- a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml +++ b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml @@ -5,7 +5,9 @@ # that there are a lot of missing installations https://github.com/EESSI/software-layer/pull/1087 . A rebuild PR like # this will have the convenient side effect of filling all those holes easyconfigs: - - CUDA-12.1.1.eb + - CUDA-12.1.1.eb: + options: + accept-eula-for: CUDA - cuDNN-8.9.2.26-CUDA-12.1.1.eb - LAMMPS-2Aug2023_update2-foss-2023a-kokkos-CUDA-12.1.1.eb - ESPResSo-4.2.2-foss-2023a-CUDA-12.1.1.eb From 8948b7fdb291ec886d362e5971d7187f72080a10 Mon Sep 17 00:00:00 2001 From: Caspar van Leeuwen Date: Thu, 21 Aug 2025 13:11:42 +0200 Subject: [PATCH 4/4] Remove CUDA, we've already rebuild. Likewise for cuDNN --- .../20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml | 4 ---- .../20250805-eb-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml | 3 --- 2 files changed, 7 deletions(-) diff --git a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml index 5d701e9eb9..0d1e7dd625 100644 --- a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml +++ b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023a-for-cuda-sanity-check.yml @@ -5,10 +5,6 @@ # that there are a lot of missing installations https://github.com/EESSI/software-layer/pull/1087 . A rebuild PR like # this will have the convenient side effect of filling all those holes easyconfigs: - - CUDA-12.1.1.eb: - options: - accept-eula-for: CUDA - - cuDNN-8.9.2.26-CUDA-12.1.1.eb - LAMMPS-2Aug2023_update2-foss-2023a-kokkos-CUDA-12.1.1.eb - ESPResSo-4.2.2-foss-2023a-CUDA-12.1.1.eb - LightGBM-4.5.0-foss-2023a-CUDA-12.1.1.eb diff --git a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml index c9f1f9b47f..155ca5141f 100644 --- a/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml +++ b/easystacks/software.eessi.io/2023.06/accel/nvidia/rebuilds/20250805-eb-5.1.1-rebuild-2023b-for-cuda-sanity-check.yml @@ -5,9 +5,6 @@ # that there are a lot of missing installations https://github.com/EESSI/software-layer/pull/1087 . A rebuild PR like # this will have the convenient side effect of filling all those holes easyconfigs: - - CUDA-12.4.0.eb: - options: - accept-eula-for: CUDA - UCX-CUDA-1.15.0-GCCcore-13.2.0-CUDA-12.4.0.eb - UCC-CUDA-1.2.0-GCCcore-13.2.0-CUDA-12.4.0.eb - OSU-Micro-Benchmarks-7.5-gompi-2023b-CUDA-12.4.0.eb