Partitions

  • There is a global list of current calculations (jobs) and calculations waiting for the requested resources.
  • A user can run 80 jobs at the same time and make up to 100 submissions.
  • The partitions, sometimes called queues, have characteristics.
  • Risk-xxx partitions have fewer limitations, but jobs running in them can be stopped by another submission made in a non-risk partition. In other words: risk-name partitions are in pre-emption mode. Do not be worry to use the risk-xxx partitions which could have been named overlimit-xxx because they allow to submit calculations beyond the “classical” limits.

Command MatriCS : digest
Command to list actived partitions : scontrol show partition | grep PartitionName
Command to view ‘PartitionName’ parameters : scontrol show partition PartitionName

shortnormalbigmembigpuquadgpunormal-amdbigmem-amdfpgaquad-a100midmemgpu
Maximum memory
per node		125 GB125 GB500 GB375 GB750 GB250 GB1000 GB1000 GB1000 GB375 GB29 GB
Maximum cores
number per node		2828284040646464644016
Maximum memory
per user		125 GB2000 GB2000 GB750 GB750 GB2000 GB2000 GB2000 GB1000 GB3000 GB29 GB
Maximum cores
number per job
and per user		2844811280405121281286432016
Optimal memory ratio
per core		4000 MB4 GB
4096 MB		16 GB
16384 MB		9 GB
9216 MB		18 GB
18432 MB		4000 MB16000 MB16000 MB16000 MB9 GB
9216 MB		1875 MB
Maximum time
per job		5 h
(10 h)		15 days (50)15 days (50)15 days15 days10 days
(15)		10 days (15)10 days (15)5 days (10)50 days (5)50 days (5)
Maximum nodes
number per job		1 (10)12 (16)3 (4)2 (3)1 (2)6 (8)1 (2)1 (2)18 (2)1
Maximum GPU
number per user		4441
Maximum FPGA
number per user		6
Partitions (risk feature in brackets)
Contents view

short and risk-short partitions

All CPU servers for quick submission.

normal and risk-normal partitions

Integrates 48 dual-processor Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz servers, 128 GB of memory.

bigmem and risk-bigmem partitions

Integrates 12 dual-processor Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz servers, 512 GB of memory.

bigpu and risk-bigpu partitions

Integrates 6 dual-processor Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz servers, 384 GB of memory and 2 Tesla V100-32G GPU cards.

quadgpu and risk-quadgpu partitions

Integrates 2 dual-processor Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz servers, 768 GB of memory and 4 Tesla V100 SXM2-32G-NVLink GPU cards.

normal-amd and risk-normal-amd partitions

Default queue. Integrates 24 dual-processor AMD EPYC 7513 @ 2.60GHz servers, 256 GB of memory.

bigmem-amd and risk-bigmem-amd partitions

Integrates 4 dual-processor AMD EPYC 7513 @ 2.60GHz servers, 1 TB of memory.

fpga and risk-fpga partitions.

Integrates 4 dual-processor AMD EPYC 7502 @ 2.50GHz servers, 1 TB of memory and 3 Xilinx U280 FPGA boards.

quad-a100 and risk-quad-a100

Integrates a dual-processor AMD EPYC 7513 @ 2.60GHz server, 1 TB of memory and 4 Tesla A100-40G GPUs.

midmem and risk-midmem partitions

midmem partition is LRCS’ ARTISTIC project dedicated. Integrates 8 dual-processor Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz servers, 384 GB of memory.

midmem and risk-midmem partitions

gpu partition is LAMFA dedicated. Integrates 1 dual-processor Intel(R) Xeon(R) Silver 4309Y CPU @ 2.80GHz servers, 384 GB of memory and 1 Tesla T4 GPU.

magma partition (priority to magma software)

Integrates an Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz dual-processor server, 512 GB of memory.
Note: For licensing reasons, the use of the magma software requires selecting the magma partition and non-magma jobs on bignode12 are subject to cancellation if a magma job request arises.
Maximum time per job : 50 days and maximum number of nodes per job : 1.

visugpu partition

Integrates an Intel(R) Xeon(R) Gold 6150 CPU @ 2.70GHz dual-processor server, 192 GB of memory and a Tesla P40 GPU card

  • Maximum requested memory : 187 GB
  • Number of cores : 36
  • Optimal memory ratio per core : 5 GB (5120 MB)
  • Maximum time per job : 1 day