Sherlock changelog

Keep up to date with software updates

2021-10-29T17:29:35.955Z

To help users stay on top of software changes on Sherlock, we’ve recently introduced a new software updates RSS feed.

It’s available from the Sherlock software list page, and you can directly add it to your RSS reader of choice.

And if you’re already following our Slack channels, you can subscribe to the new #sherlock-software-updates channel to get automatically notified of new software updates on Sherlock, directly from Slack.

Happy computing! And as usual, please feel free to [email protected]" rel="noopener" target="_blank">reach out if you have comments or questions.

You can now subscribe to Sherlock news and updates

2018-09-07T22:28:00.001Z

You can now subscribe to Sherlock news and updates, and automatically receive an email notification when a new post is published here.

To subscribe to Sherlock updates, follow the steps below:

go to https://news.sherlock.stanford.edu
click on the “Subscribe to Updates” button in the upper right corner,
enter your email address and click Subscribe.

Tensorflow 1.7.0 now available

2018-03-30T18:15:00.001Z

Tensorflow 1.7.0, which features TensorRT integration, is now available on Sherlock 2.0.

You can load it on Sherlock with:

$ ml py-tensorflow/1.7.0_py27

NB: As with all the other Tensorflow modules on Sherlock, Tensorflow 1.7.0 requires a GPU to run and can only be used in jobs submitted to GPU nodes.

For a complete changelog, please see the TensorFlow release notes.

Sherlock 2.0 limits increased

2018-03-26T20:52:59Z

We started migrating compute nodes from Sherlock 1.0 to Sherlock 2.0, as part of the 2nd phase of the transition process described in the Sherlock 2.0 transition guide.

That means that more nodes are now available on Sherlock 2.0, in both owners and public partitions (normal). As a consequence, less nodes are available to run jobs on Sherlock 1.0.

As a result, the job submission limits on Sherlock 2.0 have been increased, and are now as follows (for the normal QOS):

256 CPU-cores per user (up from 128),
512 CPU-cores per group (up from 256).

The transition process will continue, and we will gradually move more Sherlock 1.0 nodes over, further expanding Sherlock 2.0 capacity and reducing Sherlock 1.0's.

Owner nodes on Sherlock 1.0 will be migrated, in coordination with each owner group, starting April 9th. The Sherlock 1.0 environment will likely be retired over the summer.

Don't hesitate to contact us at [email protected] if you have any question.

Slurm version update

2017-08-15T21:08:20Z

Slurm has been updated to version 17.02.7.

Full changelog is available here

Slurm update

2017-06-22T22:42:14Z

Slurm has been updated to version 17.02.5. The full changelog is available here.

Of particular interest is a fix for a problem that affected users submitting jobs to multiple partitions on Sherlock 2.0, and who could receive errors about violating accounting or QOS policies.

Slurm update

2017-06-07T21:17:00Z

Slurm has been updated to version 17.02.4.

Full changelog is available here

New Slurm version

2017-04-07T00:19:45Z

Slurm has been updated to version 17.02.2.

Main changes include:

SSH sessions to the nodes while a job is running will now be correctly inherit the job limits
quicker scheduling of interactive jobs
support for PMIx job launcher

Full changelog is available here