QoS Levels and Limits¶
Each user can submit multiple jobs to the queue. The SLURM scheduler will run them in the order that they are received or as the appropriate amount of resources become available for the requested amount of time. However, to prevent the cluster from being consumed by a handful of users, some restrictions have been placed on how many resources each user can utilize at any given time. These restrictions are given in the table below.
QoS |
MaxCPUs |
MaxNodes |
MaxTime |
Priority |
Notes |
|---|---|---|---|---|---|
normal |
- |
10 |
48 hours |
20 |
Default QoS |
priority |
- |
UNLIMITED |
UNLIMITED |
50 |
By request and approval only |
special |
- |
UNLIMITED |
8 hours |
20 |
By request and approval only |
debug |
- |
500 |
30 minutes |
30 |
Debugging Jobs |
batch |
- |
500 |
8 hours |
20 |
NOAA Default |
windfall |
- |
500 |
8 hours |
1 |
NOAA Low Priority |
urgent |
- |
UNLIMITED |
8 hours |
40 |
NOAA High Priority |
ood |
2 |
1 |
8 days |
1 |
Low-Resource, Long-Running for Interactive apps |
novel |
- |
UNLIMITED |
8 hours |
50 |
|
funded |
800 |
10 |
48 hours |
20 |
|
special-nf |
1600 |
20 |
48 hours |
20 |
|
long |
- |
250 |
16 hours |
20 |
|
nofund |
800 |
10 |
2 days |
1 |
QoS |
MaxCPUs |
MaxNodes |
MaxTime |
Priority |
Notes |
|---|---|---|---|---|---|
normal |
- |
5 |
48 hours |
20 |
Default QoS |
priority |
- |
UNLIMITED |
UNLIMITED |
50 |
By request and approval only |
special |
- |
UNLIMITED |
UNLIMITED |
20 |
By request and approval only |
debug |
- |
250 |
30 minutes |
30 |
Debugging Jobs |
batch |
- |
250 |
8 hours |
20 |
NOAA Default |
windfall |
- |
250 |
8 hours |
1 |
NOAA Low Priority |
urgent |
- |
250 |
8 hours |
40 |
NOAA High Priority |
ood |
2 |
1 |
8 days |
1 |
Low-Resource, Long-Running for Interactive apps |
cadre |
20000 |
UNLIMITED |
20 minutes |
20 |
|
novel |
- |
UNLIMITED |
8 hours |
50 |
|
long |
- |
125 |
16 hours |
20 |
QoS |
MaxCPUs |
MaxNodes |
GPUs |
MaxTime |
Priority |
Notes |
|---|---|---|---|---|---|---|
normal |
128 |
1 |
A100=4, A100_1g.10gb=14, A100_3g.40gb=8 |
48 hours |
20 |
Default QoS |
special |
- |
UNLIMITED |
- |
10 days |
30 |
By request and approval only |
special-abe |
384 |
1 |
A100_1g.10gb=168 |
2 days |
20 |
|
special-me |
256 |
1 |
A100_1g.10gb=112 |
2 days |
20 |
|
class-xxxxxxx |
32 |
1 |
A100_1g.10gb=4 |
12 hours |
20 |
Class accounts have these limits unless noted |
Class account limits may vary. These limits are set in conjunction with each instructor’s needs.
QoS |
MaxCPUs |
MaxNodes |
GPUs |
MaxTime |
Priority |
Notes |
|---|---|---|---|---|---|---|
normal |
1920 |
40 |
- |
14 days |
20 |
Default QoS |
debug |
288 |
3 |
- |
30 minutes |
30 |
Debugging Jobs |
special |
- |
UNLIMITED |
- |
UNLIMITED |
20 |
By request and approval only |
priority |
- |
UNLIMITED |
- |
UNLIMITED |
100 |
By request and approval only |
ood |
2 |
1 |
- |
8 days |
1 |
Low-Resource, Long-Running for Interactive apps |
bigmem |
192 |
4 |
- |
14 days |
20 |
|
v100 |
48 |
1 |
V100=2 |
2 days |
20 |
|
mig7 |
64 |
UNLIMITED |
A100_1g.10gb=28 |
2 days |
20 |
|
a100 |
276480 |
UNLIMITED |
A100=17280 |
7 days |
20 |
|
l40s |
384 |
6 |
L40S=24 |
2 days |
20 |
|
sandbox |
- |
1 |
- |
2 hours |
10 |
For users with no association |
Note
CPUs and GPUs for A100s are in Resource Minutes. Users can request less resources for more time or more resources for less time.
QoS |
MaxCPUs |
MaxNodes |
GPUs |
MaxTime |
Priority |
Notes |
|---|---|---|---|---|---|---|
normal |
320 |
5 |
A100=2 |
48 hours |
20 |
Default QoS |
debug |
- |
2 |
- |
30 minutes |
50 |
Debugging Jobs |
special |
- |
UNLIMITED |
- |
10 days |
30 |
By request and approval only |
priority |
- |
UNLIMITED |
- |
UNLIMITED |
50 |
By request and approval only |
partner-me |
128 |
2 |
- |
48 hours |
20 |
|
partner-ngi |
128 |
1 |
A100=8 |
5 days |
20 |
|
partner-cavs |
128 |
1 |
A100=8 |
48 hours |
20 |
|
partner-igbb |
- |
1 |
- |
48 hours |
20 |
Important
Partner-xxx QoS’ can Pre-Empt jobs of other QoS when cluster utilization is high due to hardware purchase. You can learn more about Pre-Emption Here