Sherlock is a high-performance shared computing cluster for research at Stanford.
Documentation is available here.
The official way to get support with Sherlock is to email srcc-support@stanford.edu.
The #sherlock-announce and #sherlock-users Slack channels (accessible from any Slack workspace within the Stanford University organization) are also great places to receive updates about the system and to get help from research computing staff or other users.
The SRCC also runs office hours over Zoom since the COVID-19 pandemic. Please see here for details.
To request an account on Sherlock, the sponsoring Stanford faculty member should email srcc-support@stanford.edu, specifying the names and SUNet IDs of their research team members needing an account. Risa then needs to approve access to KIPAC resources.
KIPAC resources at Sherlock:
# sinfo $args -e -o "%.9R %.8m %.5D %.5c %N"
PARTITION MEMORY NODES CPUS NODELIST
kipac 191000 28 24 sh02-03n[45-72]
kipac 256000 84 32 sh03-03n[01-72],sh03-04n[01-12]
kipac 1536000 16 256 sh04-11n[13-16],sh04-12n[13-16],sh04-13n[13-16],sh04-14n[13-16]
The last row above are the 16 recently acquired CLASS:SH4_CSCALE nodes with 1.5TB RAM and 256 CPU cores each.
KIPAC members have access to a number of other partition queues as well:
$ sh_part
partition || nodes | CPU cores | GPUs || job runtime | mem/core | per-node
name public || idle total | idle total queued | idle total queued || default maximum | default maximum | cores mem(GB) gpus
-----------------------------------------------------------------------------------------------------------------------------------------------------
normal* yes || 1 211 | 747 5704 53812 | 0 0 0 || 2h 7d | 6GB 8GB | 20-64 128-384 0
bigmem yes || 1 11 | 659 824 240 | 0 0 0 || 2h 1d | 6GB 64GB | 24-256 384-4096 0
gpu yes || 0 33 | 250 1068 1925 | 25 136 220 || 1h 2d | 8GB 32GB | 20-64 191-2048 3-8
dev yes || 0 4 | 69 104 1 | 62 64 0 || 1h 2h | 6GB 8GB | 20-32 128-256 0-32
-----------------------------------------------------------------------------------------------------------------------------------------------------
kipac no || 0 128 | 3 7456 0 | 0 0 0 || 2h 7d | 6GB 8GB | 24-256 191-1536 0
hns no || 0 144 | 477 5184 5008 | 4 12 2 || 2h 7d | 6GB 26GB | 20-256 128-1536 0-4
-----------------------------------------------------------------------------------------------------------------------------------------------------
owners no || 11 1750 | 3838 63336 35309 | 514 872 2054 || 2h 2d | 4GB 48GB | 20-256 128-4096 0-8
-----------------------------------------------------------------------------------------------------------------------------------------------------
The kipac queue will likely be your workhorse. You may find the bigmem and gpu nodes useful for memory intensive applications or applications requiring GPUs. The owners and hns queues may also be useful if other queues are particularly busy (note that hns now contains 4 GPUs). See the Sherlock documentation for more information.
KIPAC Oak storage is also accessible from Sherlock; please see the Oak documentation for more information.
Sherlock has Globus access through the data transfer node (DTN). See this page for documentation. You can also use standard tools like bbcp, scp and rsync.
To get you started, here is an example using scp. Suppose your SUNet ID is alice and you want to transfer the file data.zip in your current directory to a directory named project in your home directory on Sherlock. The scp command to carry this out would be
scp data.zip alice@dtn.sherlock.stanford.edu:~/project/
Most importantly, you should not make (any, but particularly large) data transfers through the login endpoints. They are slower than through the DTN (data transfer node) and will disconnect after some timeout period. It is also more convenient to transfer through the DTN, as it (at the time of this writing) does not require duo mobile 2fa.