cc-examples/fau-systems/job-archive/cluster-tinyfat.json

466 lines
16 KiB
JSON
Raw Permalink Normal View History

2023-06-13 07:26:59 +02:00
{
"name": "tinyfat",
"metricConfig": [
{
"name": "cpu_load",
"unit": {
"base": ""
},
"scope": "node",
"aggregation": "avg",
"timestep": 60,
"peak": 128,
"normal": 64,
"caution": 60,
"alert": 20,
"subClusters": [
{
"name": "broadwell_256gb",
"peak": 24,
"normal": 24,
"caution": 12,
"alert": 10
},
{
"name": "broadwell_512gb",
"peak": 56,
"normal": 56,
"caution": 28,
"alert": 20
}
]
},
{
"name": "cpu_user",
"unit": {
"base": ""
},
"scope": "hwthread",
"aggregation": "avg",
"timestep": 60,
"peak": 100,
"normal": 50,
"caution": 20,
"alert": 10
},
{
"name": "mem_used",
"unit": {
"base": "B",
"prefix": "G"
},
"scope": "node",
"aggregation": "sum",
"timestep": 60,
"peak": 512,
"normal": 256,
"caution": 480,
"alert": 500,
"subClusters": [
{
"name": "broadwell_256gb",
"peak": 256,
"normal": 128,
"caution": 200,
"alert": 240
},
{
"name": "broadwell_512gb",
"peak": 512,
"normal": 256,
"caution": 480,
"alert": 500
}
]
},
{
"name": "flops_any",
"unit": {
"base": "Flops/s",
"prefix": "G"
},
"scope": "hwthread",
"aggregation": "sum",
"timestep": 60,
"peak": 2560,
"normal": 800,
"caution": 100,
"alert": 20,
"subClusters": [
{
"name": "broadwell_256gb",
"peak": 653,
"normal": 200,
"caution": 50,
"alert": 10
},
{
"name": "broadwell_512gb",
"peak": 1075,
"normal": 500,
"caution": 80,
"alert": 20
}
]
},
{
"name": "flops_sp",
"unit": {
"base": "Flops/s",
"prefix": "G"
},
"scope": "hwthread",
"aggregation": "sum",
"timestep": 60,
"peak": 5600,
"normal": 1000,
"caution": 200,
"alert": 50,
"subClusters": [
{
"name": "broadwell_256gb",
"peak": 653,
"normal": 200,
"caution": 50,
"alert": 10
},
{
"name": "broadwell_512gb",
"peak": 1075,
"normal": 500,
"caution": 80,
"alert": 20
}
]
},
{
"name": "flops_dp",
"unit": {
"base": "Flops/s",
"prefix": "G"
},
"scope": "hwthread",
"aggregation": "sum",
"timestep": 60,
"peak": 2300,
"normal": 500,
"caution": 100,
"alert": 50,
"subClusters": [
{
"name": "broadwell_256gb",
"peak": 325,
"normal": 100,
"caution": 25,
"alert": 10
},
{
"name": "broadwell_512gb",
"peak": 500,
"normal": 250,
"caution": 40,
"alert": 10
}
]
},
{
"name": "mem_bw",
"unit": {
"base": "B/s",
"prefix": "G"
},
"scope": "socket",
"aggregation": "sum",
"timestep": 60,
"peak": 278,
"normal": 100,
"caution": 50,
"alert": 10,
"subClusters": [
{
"name": "broadwell_256gb",
"peak": 114,
"normal": 50,
"caution": 25,
"alert": 10
},
{
"name": "broadwell_512gb",
"peak": 128,
"normal": 50,
"caution": 25,
"alert": 10
}
]
},
{
"name": "clock",
"unit": {
"base": "Hz",
"prefix": "M"
},
"scope": "hwthread",
"aggregation": "avg",
"timestep": 60,
"peak": 3000,
"normal": 2500,
"caution": 1800,
"alert": 1200,
"subClusters": [
{
"name": "broadwell_256gb",
"peak": 3800,
"normal": 3400,
"caution": 2000,
"alert": 1200
},
{
"name": "broadwell_512gb",
"peak": 3000,
"normal": 2400,
"caution": 1800,
"alert": 1200
}
]
},
{
"name": "cpu_power",
"unit": {
"base": "W"
},
"scope": "socket",
"aggregation": "sum",
"timestep": 60,
"peak": 500,
"normal": 250,
"caution": 100,
"alert": 50
},
{
"name": "mem_power",
"unit": {
"base": "W"
},
"scope": "socket",
"aggregation": "sum",
"timestep": 60,
"peak": 100,
"normal": 50,
"caution": 20,
"alert": 10
},
{
"name": "ipc",
"unit": {
"base": "IPC"
},
"scope": "hwthread",
"aggregation": "avg",
"timestep": 60,
"peak": 4,
"normal": 2,
"caution": 1,
"alert": 0.5
},
{
"name": "vectorization_ratio",
"unit": {
"base": ""
},
"scope": "hwthread",
"aggregation": "avg",
"timestep": 60,
"peak": 100,
"normal": 60,
"caution": 40,
"alert": 10
},
{
"name": "nfs4_read",
"unit": {
"base": "B/s",
"prefix": "M"
},
"scope": "node",
"aggregation": "sum",
"timestep": 60,
"peak": 6,
"normal": 4,
"caution": 2,
"alert": 1
},
{
"name": "nfs4_write",
"unit": {
"base": "B/s",
"prefix": "M"
},
"scope": "node",
"aggregation": "sum",
"timestep": 60,
"peak": 6,
"normal": 4,
"caution": 2,
"alert": 1
},
{
"name": "nfs4_total",
"unit": {
"base": "B/s",
"prefix": "M"
},
"scope": "node",
"aggregation": "sum",
"timestep": 60,
"peak": 6,
"normal": 4,
"caution": 2,
"alert": 1
}
],
"subClusters": [
{
"name": "broadwell_512gb",
"nodes": "tf040,tf041,tf042",
"processorType": "Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz",
"socketsPerNode": 2,
"coresPerSocket": 14,
"threadsPerCore": 2,
"flopRateScalar": {
"unit": {
"base": "F/s",
"prefix": "G"
},
"value": 158
},
"flopRateSimd": {
"unit": {
"base": "F/s",
"prefix": "G"
},
"value": 1236
},
"memoryBandwidth": {
"unit": {
"base": "B/s",
"prefix": "G"
},
"value": 128
},
"topology": {
"node": [
0, 28, 1, 29, 2, 30, 3, 31, 4, 32, 5, 33, 6, 34, 7, 35, 8, 36, 9, 37, 10, 38, 11, 39, 12, 40, 13, 41, 14, 42, 15, 43, 16, 44, 17, 45, 18, 46, 19, 47, 20, 48, 21, 49, 22, 50, 23, 51, 24, 52, 25, 53, 26, 54, 27, 55
],
"socket": [
[ 0, 28, 1, 29, 2, 30, 3, 31, 4, 32, 5, 33, 6, 34, 7, 35, 8, 36, 9, 37, 10, 38, 11, 39, 12, 40, 13, 41 ],
[ 14, 42, 15, 43, 16, 44, 17, 45, 18, 46, 19, 47, 20, 48, 21, 49, 22, 50, 23, 51, 24, 52, 25, 53, 26, 54, 27, 55 ]
],
"memoryDomain": [
[ 0, 28, 1, 29, 2, 30, 3, 31, 4, 32, 5, 33, 6, 34 ],
[ 7, 35, 8, 36, 9, 37, 10, 38, 11, 39, 12, 40, 13, 41 ],
[ 14, 42, 15, 43, 16, 44, 17, 45, 18, 46, 19, 47, 20, 48 ],
[ 21, 49, 22, 50, 23, 51, 24, 52, 25, 53, 26, 54, 27, 55 ]
],
"core": [
[ 0, 28 ], [ 1, 29 ], [ 2, 30 ], [ 3, 31 ], [ 4, 32 ], [ 5, 33 ], [ 6, 34 ], [ 7, 35 ], [ 8, 36 ], [ 9, 37 ], [ 10, 38 ], [ 11, 39 ], [ 12, 40 ], [ 13, 41 ], [ 14, 42 ], [ 15, 43 ], [ 16, 44 ], [ 17, 45 ], [ 18, 46 ], [ 19, 47 ], [ 20, 48 ], [ 21, 49 ], [ 22, 50 ], [ 23, 51 ], [ 24, 52 ], [ 25, 53 ], [ 26, 54 ], [ 27, 55 ]
]
}
},
{
"name": "broadwell_256gb",
"nodes": "tf0[50-57]",
"processorType": "Intel(R) Xeon(R) CPU E5-2643 v4 @ 3.40GHz",
"socketsPerNode": 2,
"coresPerSocket": 6,
"threadsPerCore": 2,
"flopRateScalar": {
"unit": {
"base": "F/s",
"prefix": "G"
},
"value": 85
},
"flopRateSimd": {
"unit": {
"base": "F/s",
"prefix": "G"
},
"value": 672
},
"memoryBandwidth": {
"unit": {
"base": "B/s",
"prefix": "G"
},
"value": 114
},
"topology": {
"node": [
0, 12, 1, 13, 2, 14, 3, 15, 4, 16, 5, 17, 6, 18, 7, 19, 8, 20, 9, 21, 10, 22, 11, 23
],
"socket": [
[ 0, 12, 1, 13, 2, 14, 3, 15, 4, 16, 5, 17 ],
[ 6, 18, 7, 19, 8, 20, 9, 21, 10, 22, 11, 23 ]
],
"memoryDomain": [
[ 0, 12, 1, 13, 2, 14, 3, 15, 4, 16, 5, 17 ],
[ 6, 18, 7, 19, 8, 20, 9, 21, 10, 22, 11, 23 ]
],
"core": [
[ 0, 12 ], [ 1, 13 ], [ 2, 14 ], [ 3, 15 ], [ 4, 16 ], [ 5, 17 ], [ 6, 18 ], [ 7, 19 ], [ 8, 20 ], [ 9, 21 ], [ 10, 22 ], [ 11, 23 ]
]
}
},
{
"name": "rome_512gb",
"nodes": "tf0[60-95]",
"processorType": "AMD EPYC 7502 32-Core Processor ",
"socketsPerNode": 2,
"coresPerSocket": 32,
"threadsPerCore": 2,
"flopRateScalar": {
"unit": {
"base": "F/s",
"prefix": "G"
},
"value": 553
},
"flopRateSimd": {
"unit": {
"base": "F/s",
"prefix": "G"
},
"value": 3198
},
"memoryBandwidth": {
"unit": {
"base": "B/s",
"prefix": "G"
},
"value": 278
},
"topology": {
"node": [
0, 64, 1, 65, 2, 66, 3, 67, 4, 68, 5, 69, 6, 70, 7, 71, 8, 72, 9, 73, 10, 74, 11, 75, 12, 76, 13, 77, 14, 78, 15, 79, 16, 80, 17, 81, 18, 82, 19, 83, 20, 84, 21, 85, 22, 86, 23, 87, 24, 88, 25, 89, 26, 90, 27, 91, 28, 92, 29, 93, 30, 94, 31, 95, 32, 96, 33, 97, 34, 98, 35, 99, 36, 100, 37, 101, 38, 102, 39, 103, 40, 104, 41, 105, 42, 106, 43, 107, 44, 108, 45, 109, 46, 110, 47, 111, 48, 112, 49, 113, 50, 114, 51, 115, 52, 116, 53, 117, 54, 118, 55, 119, 56, 120, 57, 121, 58, 122, 59, 123, 60, 124, 61, 125, 62, 126, 63, 127
],
"socket": [
[ 0, 64, 1, 65, 2, 66, 3, 67, 4, 68, 5, 69, 6, 70, 7, 71, 8, 72, 9, 73, 10, 74, 11, 75, 12, 76, 13, 77, 14, 78, 15, 79, 16, 80, 17, 81, 18, 82, 19, 83, 20, 84, 21, 85, 22, 86, 23, 87, 24, 88, 25, 89, 26, 90, 27, 91, 28, 92, 29, 93, 30, 94, 31, 95 ],
[ 32, 96, 33, 97, 34, 98, 35, 99, 36, 100, 37, 101, 38, 102, 39, 103, 40, 104, 41, 105, 42, 106, 43, 107, 44, 108, 45, 109, 46, 110, 47, 111, 48, 112, 49, 113, 50, 114, 51, 115, 52, 116, 53, 117, 54, 118, 55, 119, 56, 120, 57, 121, 58, 122, 59, 123, 60, 124, 61, 125, 62, 126, 63, 127 ]
],
"memoryDomain": [
[ 0, 64, 1, 65, 2, 66, 3, 67, 4, 68, 5, 69, 6, 70, 7, 71 ],
[ 8, 72, 9, 73, 10, 74, 11, 75, 12, 76, 13, 77, 14, 78, 15, 79 ],
[ 16, 80, 17, 81, 18, 82, 19, 83, 20, 84, 21, 85, 22, 86, 23, 87 ],
[ 24, 88, 25, 89, 26, 90, 27, 91, 28, 92, 29, 93, 30, 94, 31, 95 ],
[ 32, 96, 33, 97, 34, 98, 35, 99, 36, 100, 37, 101, 38, 102, 39, 103 ],
[ 40, 104, 41, 105, 42, 106, 43, 107, 44, 108, 45, 109, 46, 110, 47, 111 ],
[ 48, 112, 49, 113, 50, 114, 51, 115, 52, 116, 53, 117, 54, 118, 55, 119 ],
[ 56, 120, 57, 121, 58, 122, 59, 123, 60, 124, 61, 125, 62, 126, 63, 127 ]
],
"core": [
[ 0, 64 ], [ 1, 65 ], [ 2, 66 ], [ 3, 67 ], [ 4, 68 ], [ 5, 69 ], [ 6, 70 ], [ 7, 71 ], [ 8, 72 ], [ 9, 73 ], [ 10, 74 ], [ 11, 75 ], [ 12, 76 ], [ 13, 77 ], [ 14, 78 ], [ 15, 79 ], [ 16, 80 ], [ 17, 81 ], [ 18, 82 ], [ 19, 83 ], [ 20, 84 ], [ 21, 85 ], [ 22, 86 ], [ 23, 87 ], [ 24, 88 ], [ 25, 89 ], [ 26, 90 ], [ 27, 91 ], [ 28, 92 ], [ 29, 93 ], [ 30, 94 ], [ 31, 95 ], [ 32, 96 ], [ 33, 97 ], [ 34, 98 ], [ 35, 99 ], [ 36, 100 ], [ 37, 101 ], [ 38, 102 ], [ 39, 103 ], [ 40, 104 ], [ 41, 105 ], [ 42, 106 ], [ 43, 107 ], [ 44, 108 ], [ 45, 109 ], [ 46, 110 ], [ 47, 111 ], [ 48, 112 ], [ 49, 113 ], [ 50, 114 ], [ 51, 115 ], [ 52, 116 ], [ 53, 117 ], [ 54, 118 ], [ 55, 119 ], [ 56, 120 ], [ 57, 121 ], [ 58, 122 ], [ 59, 123 ], [ 60, 124 ], [ 61, 125 ], [ 62, 126 ], [ 63, 127 ]
]
}
}
]
}