PROXIMA
| Hardware Characteristics
|
Proxima
|
| Type
|
PC cluster
|
| Architecture
|
Intel Xeon Platinum 8480+
|
| Network interfaces
|
InfiniBand EDR/NDR, Ethernet 10,25,100Gbit
|
| Internal connection network
|
Infiniband EDR/NDR
|
| Number of GPU cards
|
348
|
| Total computational power
|
23.7 PFLOPS
|
| System memory size
|
66.8 TB
|
| Disk memory capacity
|
|
ALTAIR
| Hardware Characteristics
|
Altair
|
| Type
|
PC cluster
|
| Architecture
|
Intel Xeon Platinum 8268
|
| Network interfaces
|
InfiniBand EDR/NDR, Ethernet 10,25,100Gbit
|
| Internal connection network
|
Infiniband EDR/NDR
|
| Number of CPU cores
|
63360
|
| Total computational power
|
5.9 PFLOPS
|
| System memory size
|
300 TB
|
| Disk memory capacity
|
|
Proxima CPU Nodes
| Node commercial name
|
CPU model
|
Number of nodes
|
Number of processors and cores
|
RAM amount per node
|
Node computational power
|
Node class tag
|
| Hewlett Packard Enterprise
|
Intel Xeon Platinum 8480+
|
60
|
2x56
|
384 GB
|
7 TFLOPS
|
intel,
|
Proxima GPU Nodes
| Node commercial name
|
CPU model
|
Number of nodes
|
Number of processors and cores
|
RAM amount per node
|
GPU model
|
Number of graphics cards per node
|
GPU node computational power
|
Node class tag
|
| HPE Cray XD665
|
AMD EPYC 9334
|
87
|
2x32
|
768 GB
|
Nvidia H100-94 SXM5, 94GB HBM2e
|
4
|
268 TFLOPS
|
|
Altair CPU Nodes
| Node commercial name
|
CPU model
|
Number of nodes
|
Number of processors and cores
|
RAM amount per node
|
Node computational power
|
Node class tag
|
| HUAWEI CH121 V5
|
Intel Xeon Platinum 8268
|
1320
|
2 x 24
|
192/384 GB
|
4.4 TFLOPS
|
intel, cascade
|
Eagle GPU Nodes
| Node commercial name
|
CPU model
|
Number of nodes
|
Number of processors and cores
|
RAM amount per node
|
GPU model
|
Number of graphics cards per node
|
GPU node computational power
|
Node class tag
|
|
|
Xeon Gold Gold 5115
|
3
|
2x10
|
92 GB
|
Nvidia V100 32GB
|
2
|
15.6 TFLOPS
|
|
Altair GPU Nodes
| Node commercial name
|
CPU model
|
Number of nodes
|
Number of processors and cores
|
RAM amount per node
|
GPU model
|
Number of graphics cards per node
|
GPU node computational power
|
Node class tag
|
|
|
Intel Xeon 6242
|
9
|
2x16
|
384 GB
|
Nvidia V100 32GB
|
8
|
62.4 TFLOPS
|
|
| Other Information
|
| Operating System
|
| Type
|
GNU Linux
|
| Computer domain name
|
eagle.man.poznan.pl
|
| Batch job management system
|
SLURM
|
Queues
Running calculations through the SLURM system is done within queues that differ in priority, limits, or access rights (and sometimes also in available resources)
We have the following queues:
| Queue Name
|
Job time limit
|
Default job time limit
|
Standard RAM amount/core
|
Notes
|
| standard
|
7 days
|
1 day
|
2 GB
|
Default queue, can run both parallel and sequential tasks
|
| fast
|
1 hour
|
10 minutes
|
-
|
high priority; maximum number of nodes 1
|
| tesla
|
7 days
|
1 day
|
2 GB
|
Queue with GPU nodes
|
Altair Queues
| Altair Queue Name
|
Job time limit
|
Default job time limit
|
Standard RAM amount/core
|
Notes
|
| altair
|
7 days
|
1 day
|
2 GB
|
Default queue, can run both parallel and sequential tasks
|
|
| altair-gpu
|
7 days
|
1 day
|
2 GB
|
Queue with GPU nodes
|
