ibm server stock imageHere we present the results of our benchmarking of Bede. The benchmarks we are running are our HECBioSim benchmark suite here. Our benchmarks aren't too heavily tuned as these are designed to be used by our community to work out how much HPC time to ask for on this resource, so are set at a level that would be reasonable for any normal biomolecular MD simulation. More information on the benchmarks themselves can be found at the above link.

Bede is one of the UK Tier2 facilities and is comprised of 32 main GPU nodes each of which is equipped with two IBM Power9 CPUs with 512GB of RAM per node. Each node has four NVidia V100 32GB GPUs and is connected to a high performance Lustre parallel filesystem, internode communication is handled by a high speed and low latency infiniband interconnect.


AMBER 18 - single node, single GPU
20k atom - 346 ns/day
61k atom - 217 ns/day
465k atom - 33 ns/day
1400k atom - 10 ns/day
3000k atom - 4.44 ns/day

AMBER 20 - single node, single GPU
20k atom - 342 ns/day
61k atom - 211 ns/day
465k atom - 32 ns/day
1400k atom - 10 ns/day
3000k atom - 4.56 ns/day

Gromacs 2020.3 - single node, single GPU
20k atom - 414 ns/day
61k atom - 164 ns/day
465k atom - 25 ns/day
1400k atom - 7.8 ns/day
3000k atom - 4.2 ns/day

Gromacs 2022.2 - single node, single GPU
20k atom - 464 ns/day
61k atom - 178 ns/day
465k atom - 27 ns/day
1400k atom - 9.0 ns/day
3000k atom - 4.9 ns/day

NAMD 2.14 - single node, single GPU
20k atom - 49 ns/day
61k atom - 21 ns/day
465k atom - 2.3 ns/day
1400k atom - 0.79 ns/day
3000k atom - 0.41 ns/day

NAMD 3-alpha12 - single node, single GPU - do note, this is an alpha release of 3.0
20k atom - 159 ns/day
61k atom - 130 ns/day
465k atom - 15 ns/day
1400k atom - 5.2 ns/day
3000k atom - 2.3 ns/day


In our initial benchmarking run we focused on single node with single GPUs on Bede. We will be adding more stable codes and starting to look at how to get the best out of single node multiple GPU and also multinode GPU simulations soon.