Scheduling

Showing 52 of 16395

GSLA Resource Loaning - Add the ability to loan to specific users or projects

When a GSLA resource pool is defined for a specific type of workload, it is not practical or desirable to loan to just any other type of job profile. The work-a-round today is to include a resource limit definition to prevent this from occurring. ...

22 days ago in Spectrum LSF / Scheduling 0 Under review

1 VOTE

Set different requeue exit values for different hosts or hostgroups

Sometimes it would be helpful to set different requeue exit values for different hostgroups, something like (i.e. in lsb.applications):REQUEUE_EXIT_VALUES 138 linux_hosts 210 windows_hosts 127 which would trigger a requeue for jobs exiting with va...

28 days ago in Spectrum LSF / Scheduling 0 Under review

1 VOTE

Increase amount of characters in job description

This enhancement would enable us to store additional job attributes within each job record, such as specific flow information. By enriching the job metadata, machine learning algorithms can leverage this information to make more informed predictio...

29 days ago in Spectrum LSF / Scheduling 0 Under review

3 VOTE

Increase available number of global job ids for LSF

Current job id is limited to 32b. With ever growing number of jobs and scaling of multi-cluster use, the job id should be extended beyond 32b to avoid frequent rollover situations and/or cases of redundant ids. Recommend to increase to 64b or more.

4 months ago in Spectrum LSF / Scheduling 0 Planned for future release

3 VOTE

Affinity scheduling must take L3 cache hierarchy into account

Newer CPUs have multiple L3 caches, with dedicated cores per L3 cache. Running multi-threaded jobs on all cores within a shared L3 cache can reduce runtime in our cases up 70%, compared to running over multiple L3 caches. Think of the AMD EPYC 935...

8 months ago in Spectrum LSF / Scheduling 2 Planned for future release

1 VOTE

let "-gpu" support "||"

This allows more than one gmodel to be chosen when submitting a job. So it is a useful feature.

3 months ago in Spectrum LSF / Scheduling 1 Functionality already exists

2 VOTE

Map memory reservation to cgroup memory.low

Mapping the memory reservation of a job to the cgroup memory.low could protect jobs from running short on memory, while in parallel allowing more memory to be used if it's available. The soft limit instead of a hard limit should prevent users to b...

8 months ago in Spectrum LSF / Scheduling 2 Future consideration

1 VOTE

Do not start jobs with less than X cores per host

It is useful for when ptiles or slots do not fulfill your needs. Using an example, we have an MPI code that requests 100 cores. We don't tell LSF how to grab those cores, and sometimes it will go out and span across 12 nodes, taking 19 cores on on...

4 months ago in Spectrum LSF / Scheduling 0 Future consideration

1 VOTE

for loadSched memory if the submitted job has memory reservation the job should not be run if the nodes mem used + job memory reservation > loadSched memory limit

because users want to keep a certain amount of memory free. And currently the limit can be bypassed.

6 months ago in Spectrum LSF / Scheduling 0 Future consideration

4 VOTE

Add possibility to assign cores from the same CCX in AMD CPUs to make use of the same L3 cache.

With a specific 2-core workload we noticed significant performance improvements of about 50% in shorter runtime when both cores reside on the same L3 cache group (CCX) of an AMD EPYC CPU.Therefore it would be very beneficial to be able to have LSF...

about 2 years ago in Spectrum LSF / Scheduling 0 Planned for future release

Please enter your email address

FILTER BY CATEGORY