blob: fae3da5535f82759b92ae6ca05c667a31bcb3a45 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
"""
Derived Quantities.
"""
def compute_utilization(number_of_nodes_down, \
number_of_jobs_by_partition):
"""
Compute Cluster Utilization.
@todo: Less Hardcoding.
@param: number_of_nodes_down - [Dict {'partition': 2, ...}]
@param: number_of_jobs_by_partition - [Dict {'zbox': {'running': 1}, ...]
@return: utilization - [Dict {'cpu': 0.3, 'tasna': 0.5, ..., 'gpu': 0.9}]
"""
utilization = {}
# CPU Utilization (Without Hyperthreading)
# 2 Sockets per Server, 8 Cores per Socket, 192 Servers = 3072 Cores
total_cpu_cores = 16*192
cpu_nodes_down = number_of_nodes_down['cpu']*16
allocated_cpu_cores = number_of_jobs_by_partition['cpu']['running']
utilization['cpu'] = \
float(allocated_cpu_cores) / float(total_cpu_cores - cpu_nodes_down)
# Tasna Utilization
# 4 GTX 590 Boards per Server, 2 GPUs per Board, 5 Servers = 40 Slots
total_tasna_slots = 40 - number_of_nodes_down['tasna']
allocated_tasna_slots = number_of_jobs_by_partition['tasna']['running']
utilization['tasna'] = \
float(allocated_tasna_slots) / float(total_tasna_slots)
# Vesta Utilization
# 8 K80 Boards per Server, 2 GPUs per Board, 2 Servers = 32 Slots
total_vesta_slots = 32 - number_of_nodes_down['vesta']
allocated_vesta_slots = number_of_jobs_by_partition['vesta']['running']
utilization['vesta'] = \
float(allocated_vesta_slots) / float(total_vesta_slots)
# GPU Utilization
total_gpu_slots = total_tasna_slots + total_vesta_slots
allocated_gpu_slots = allocated_tasna_slots + allocated_vesta_slots
utilization['gpu'] = float(allocated_gpu_slots) / float(total_gpu_slots)
return utilization
|
