Files
project_6/cccl_upstream/benchmarks/scripts/submit_benchmark_job.sh
EngineX CI 56fd68e7dd [INFRA] Import NVIDIA/CCCL upstream as optimization reference library
CCCL (CUDA C++ Core Libraries) provides:
- CUB: device/block/warp-level GPU primitives (reduce, scan, sort, topk)
- Thrust: high-level parallel algorithms (transform_reduce, sort, scan)
- libcudacxx: CUDA C++ standard library (atomics, barriers, memory)
- cudax: experimental features (memory resources, allocators)
- Tuning policies: per-SM hardware-specific algorithm parameters

Competition optimization vectors mapped to CCCL:
- Output TPS (83% weight): warp_reduce, block_reduce, device_topk
- Input TPS (14% weight): device_scan, block_load, prefetch
- Cache TPS (3% weight): prefix caching strategy patterns
- Memory (0.9 util): pooled/cached/buddy allocators

Source: https://github.com/NVIDIA/cccl (shallow clone, HEAD only)
License: Apache-2.0
2026-07-30 09:35:51 +00:00

48 lines
1.3 KiB
Bash
Executable File

#!/usr/bin/env bash
# This script schedules a SLURM job via crun on computelab to run all CCCL benchmarks and produce a benchmark database
# TODO: set those accordingly
scratch=/home/scratch."$USER"_sw
node_selector="cpu.arch=x86_64 and gpu.product_name='*B200*'"
container_image="rapidsai/devcontainers:26.06-cpp-gcc14-cuda13.2"
jobtime="4:00:00"
benchmark_preset="benchmark"
batch_script=$scratch/batch.sh
cat << BATCH_SCRIPT > "$batch_script"
#!/usr/bin/env bash
pip install --break-system-packages fpzip pandas scipy
# clone CCCL
host=\$(hostname)
cd $scratch
if [ -d "\$host/cccl" ]; then
rm -r \$host/cccl
fi
mkdir \$host
cd \$host
git clone --depth 1 git@github.com:NVIDIA/cccl.git
cd cccl
# configure cmake
mkdir build_perf
cd build_perf
cmake .. --preset $benchmark_preset
# run benchmarks
export CUDA_VISIBLE_DEVICES=0
export PYTHONPATH=../benchmarks/scripts/
../benchmarks/scripts/run.py
echo "Benchmark done. Results in $scratch/\$host/cccl/build_perf/cccl_meta_bench.db"
BATCH_SCRIPT
chmod +x "$batch_script"
# schedule SLURM job
echo "Scheduling script $batch_script"
echo "#################################################################################"
cat "$batch_script"
echo "#################################################################################"
crun -q "$node_selector" -ex -t "$jobtime" -img "$container_image" -b "$batch_script"