Overview
The GIGA cluster is a high-performance computing (HPC) system, also referred to as a cluster, provided by the GIGA to allow its members to perform computationaly intesive tasks.
It is composed of:
computing nodes to perfom analyses, hosted at the CHU (B34),
a scratch disk to store the temporary results,
master nodes to connect the different parts and manage the jobs running on the computing nodes, located at the SEGI (B26).
The GIGA cluster is directly connected to the mass storage to allow fast and easy migration of data from one to the other.
Note
The GIGA cluster is optimised for tasks that require a lot of memory. If your analysis is mainly CPU intensive, and hence requires a large number of cores, or if you need GPU accelerated nodes, you should consider using the CÉCI clusters.
Specifications
Hint
Depending on your affilition, you may have access to more partitions and/or nodes. To get more information about their respective specifications, refer to the Slurm documentation.
Here are the specifications of the default nodes and partitions available on the GIGA cluster.
Nodes
The GIGA cluster features several types of computing nodes. Here are the specifications for each of them.
Node |
Node ID |
CPU type |
CPU count[1] |
RAM |
Storage type |
Storage[2] |
---|---|---|---|---|---|---|
chugen |
001-002 |
intel |
20 |
128 GB |
SSD |
1.8 TB |
chugen |
003-004 |
intel |
24 |
128 GB |
SSD |
1.8 TB |
ptfgen |
001-004 |
intel |
16 |
256 GB |
SSD |
1.5 TB |
ptfgen |
006,009,012 |
amd |
32 |
128 GB |
SSD |
1.8 TB |
ptfgen |
007 |
amd |
32 |
128 GB |
HDD |
450 GB |
ptfgen |
013-014 |
intel |
24 |
256 GB |
SSD |
1.8 TB |
ptfgen |
015-016 |
intel |
24 |
384 GB |
SSD |
1.8 TB |
urtgen |
001-004 |
intel |
24 |
512 GB |
SSD |
1.8 TB |
urtgen |
005-008 |
intel |
40 |
192 GB |
SSD |
1.8 TB |
urtgen |
009-012 |
intel |
36 |
128 GB |
SSD |
1.8 TB |
Partitions
Warning
The default queue defq*
does not contain any node.
A job submitted without specifying a partition will remain pending forever.
Hence, it is required to set a queue when submitting a job.
Several partitions are available, each serving a different purpose.
Partition |
Time limit |
Nodes count |
Nodes list |
---|---|---|---|
all_24hrs |
1 day |
6 |
ptfgen[002-003,015-016] |
all_10days |
10 days |
3 |
ptfgen[015-016] |
all_5days |
5 days |
8 |
ptfgen[006-007,009,012-016] |
all_5hrs |
5 hours |
24 |
chugen[001-004] |
kosmos |
infinite |
4 |
ptfgen[006-007,009,012] |
nextflow |
infinite |
1 |
chugen001 |