Partition & QoS¶
Partitions¶
A partition is a set of compute nodes grouped logically based on their hardware features. The table below shows the available partitions and their properties / features in Madhava HPC
Partition | Default / Max Job duration | nodes | cores per node | RAM(GB) per node | State |
---|---|---|---|---|---|
testq | 01:00:00 | compute[01-31] | 40 | 192GB | UP |
shortq | infinite | compute[01-31] | 40 | 192GB | UP |
medium | infinite | compute[01-31] | 40 | 192GB | UP |
longq | infinite | compute[01-31] | 40 | 192GB | UP |
testgpuq | 01:00:00 | compute[32-33] | 40 | 192GB | UP |
gpu q | infinite | compute[32-33] | 40 | 192GB | UP |
Quality of Service (QoS)¶
Each QoS is assigned a set of limits to be applied to the job, dictating the limit in the resources and partitions that a job is entitled to request. The table below shows the available QoS in Madhava HPC and their allowed partitions / resources limits.
Name | Wall Time | Max Jobs Per User | Max Submit Per User | Max Nodes | Min Nodes | Max Cores | Resources |
---|---|---|---|---|---|---|---|
testq | 01:00:00 | 2 | 2 | 1 | 1 | 40 | |
shortq | infinite | 4 | 5 | 1 | 1 | 40 | |
medium | infinite | 3 | 4 | 3 | 2 | 120 | |
longq | infinite | 1 | 1 | 8 | 5 | 320 | |
testgpuq | 01:00:00 | 2 | 2 | 1 | 1 | 40 | 2 GPU |
gpu q | infinite | 1 | 2 | 1 | 1 | 40 | 2 GPU |
Note
Users are advised to specify a suitable QoS depending on the job’s requirement.
Note
For those serial jobs or multi-threaded (OpenMP) jobs that can only be executed on a single node requested to uswe "shortq", for the jobs requires more running time then the "mediumq" and “longq” QoS is a more preferable one.