Skip to content

Use rocm 7.1.1 over 7.2.1 #78

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.

Sign up for GitHub

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

Merged
PatrickRMiles merged 2 commits into from
Jun 18, 2026
Merged

Use rocm 7.1.1 over 7.2.1 #78

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
00d1704
use 7.1.1 over 7.2
michaelmckinsey1 Jun 11, 2026
0028de9
Update scaffold-tuolumne-torchpypi.job
michaelmckinsey1 Jun 18, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,7 +63,9 @@ cuda = [
"mpi4py==4.1.1",
]
rocm = [
"torch==2.12.0+rocm7.2",
"torch==2.12.0+rocm7.1",
"torchaudio==2.11.0+rocm7.1",
"torchvision==0.27.0+rocm7.1",
Comment thread
PatrickRMiles marked this conversation as resolved.
"mpi4py==4.1.1",
]
rocmwci = [
Expand Down
8 changes: 4 additions & 4 deletions scripts/install-tuolumne-torchpypi.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
ml load python/3.11.5 && python3 -m venv .venvs/scaffoldvenv-tuo-pypi && source .venvs/scaffoldvenv-tuo-pypi/bin/activate && pip install --upgrade pip
ml cce/21.0.1 cray-mpich/9.1.0 rocm/7.2.1 rccl/fast-env-slows-mpi
pip install -e .[rocm] --find-links https://download.pytorch.org/whl/torch/ --find-links https://download.pytorch.org/whl/triton-rocm/ 2>&1 | tee install.log
ml load python/3.13.2 && python3 -m venv .venvs/scaffoldvenv-tuo-pypi && source .venvs/scaffoldvenv-tuo-pypi/bin/activate && pip install --upgrade pip
ml cce/21.0.1 cray-mpich/9.1.0 rocm/7.1.1 rccl/fast-env-slows-mpi
pip install -e .[rocm] --find-links https://download.pytorch.org/whl/torch/ --find-links https://download.pytorch.org/whl/torchaudio/ --find-links https://download.pytorch.org/whl/torchvision/ --find-links https://download.pytorch.org/whl/triton-rocm/ 2>&1 | tee install.log
# libmpi.so.12 does not exist => ls /opt/cray/pe/lib64/ | grep libmpi
patchelf --replace-needed libmpi.so.12 libmpi_gnu.so.12 .venvs/scaffoldvenv-tuo-pypi/lib/python3.11/site-packages/mpi4py/MPI.mpich.cpython-311-x86_64-linux-gnu.so
patchelf --replace-needed libmpi.so.12 libmpi_gnu.so.12 .venvs/scaffoldvenv-tuo-pypi/lib/python3.13/site-packages/mpi4py/MPI.mpich.cpython-313-x86_64-linux-gnu.so
12 changes: 6 additions & 6 deletions scripts/scaffold-tuolumne-torchpypi.job
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,20 +7,20 @@
# flux: -qpdebug
# flux: -B flask

ml cce/21.0.1 cray-mpich/9.1.0 rocm/7.2.1 rccl/fast-env-slows-mpi
ml cce/21.0.1 cray-mpich/9.1.0 rocm/7.1.1 rccl/fast-env-slows-mpi

. .venvs/scaffoldvenv-tuo-pypi/bin/activate

export NCCL_NET_PLUGIN=/collab/usr/global/tools/rccl/toss_4_x86_64_ib_cray/rocm-7.2.0/install/lib/librccl-net.so
export NCCL_NET_PLUGIN=/collab/usr/global/tools/rccl/toss_4_x86_64_ib_cray/rocm-7.1.1/install/lib/librccl-net.so

# Disable direct convolution benchmarking (should speedup warmup by a significant amount, does the below three options together)
# export MIOPEN_DEBUG_CONV_DIRECT=0
export MIOPEN_DEBUG_CONV_DIRECT=0
# Disable direct naive convolution benchmarking (naive_conv_ab_nonpacked_fwd_ndhwc_half_double_half.kd)
export MIOPEN_DEBUG_CONV_DIRECT_NAIVE_CONV_FWD=0
# export MIOPEN_DEBUG_CONV_DIRECT_NAIVE_CONV_FWD=0
# Disable naive_conv_ab_nonpacked_bwd_ndhwc_half_double_half.kd
export MIOPEN_DEBUG_CONV_DIRECT_NAIVE_CONV_BWD=0
# export MIOPEN_DEBUG_CONV_DIRECT_NAIVE_CONV_BWD=0
# Disable naive_conv_ab_nonpacked_wrw_ndhwc_half_double_half.kd
export MIOPEN_DEBUG_CONV_DIRECT_NAIVE_CONV_WRW=0
# export MIOPEN_DEBUG_CONV_DIRECT_NAIVE_CONV_WRW=0

CONFIG_PATH="$(pwd)/ScaFFold/configs/benchmark_default.yml"
FRACT_BASE_DIR="${FRACT_BASE_DIR:-$(pwd)/ScaFFold/fractals}"
Expand Down
Loading