Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

use package power metrics #43

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

use package power metrics #43

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
139 changes: 72 additions & 67 deletions grafana/dashboard_gpu.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,13 +14,6 @@
"description": "",
"type": "datasource",
"pluginId": "__expr__"
},
{
"name": "VAR_G_MI300_MODEL",
"type": "constant",
"label": "MI300 Model",
"value": "102-G30211-00",
"description": ""
}
],
"__elements": {},
Expand All @@ -40,7 +33,7 @@
"type": "grafana",
"id": "grafana",
"name": "Grafana",
"version": "11.2.2+security-01"
"version": "11.2.2"
},
{
"type": "datasource",
Expand Down Expand Up @@ -108,10 +101,10 @@
"showLineNumbers": false,
"showMiniMap": false
},
"content": "#### VENDOR\n${g_card_vendor}\n\n#### SERIES\n${g_card_series}\n\n#### MODEL\n${g_card_model}\n\n#### SERIAL\n${g_serial_number}",
"content": "##### VENDOR\n${g_card_vendor}\n\n##### SERIES\n${g_card_series}\n\n##### MODEL\n${g_card_model}\n\n##### SERIAL\n${g_serial_number}\n\n##### UUID\n${g_gpu_uuid}",
"mode": "markdown"
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"targets": [
{
"datasource": {
Expand All @@ -130,7 +123,6 @@
"useBackend": false
}
],
"title": "Device",
"type": "text"
},
{
Expand All @@ -154,15 +146,15 @@
"content": "#### HOST\n${g_hostname}\n\n#### GPU ID\n${g_gpu_id}",
"mode": "markdown"
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"type": "text"
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"description": "GPU current power",
"description": "GPU package power usage",
"fieldConfig": {
"defaults": {
"color": {
Expand All @@ -178,7 +170,7 @@
},
{
"color": "red",
"value": 80
"value": 700
}
]
},
Expand Down Expand Up @@ -208,7 +200,7 @@
"showThresholdMarkers": true,
"sizing": "auto"
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"targets": [
{
"datasource": {
Expand All @@ -218,17 +210,35 @@
"disableTextWrap": false,
"editorMode": "builder",
"exemplar": false,
"expr": "gpu_package_power{gpu_uuid=\"$g_gpu_uuid\", hostname=\"$g_hostname\"}",
"expr": "gpu_average_package_power{gpu_uuid=\"$g_gpu_uuid\", hostname=\"$g_hostname\"}",
"fullMetaSearch": false,
"hide": false,
"includeNullMetadata": true,
"instant": false,
"legendFormat": "{{hostname}}[{{gpu_id}}]",
"range": true,
"refId": "A",
"useBackend": false
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"disableTextWrap": false,
"editorMode": "builder",
"expr": "gpu_package_power{gpu_uuid=\"$g_gpu_uuid\", hostname=\"$g_hostname\"}",
"fullMetaSearch": false,
"hide": false,
"includeNullMetadata": true,
"instant": false,
"legendFormat": "{{hostname}}[{{gpu_id}}]",
"range": true,
"refId": "B",
"useBackend": false
}
],
"title": "GPU Package Power",
"title": "GPU Power Usage",
"transformations": [
{
"id": "calculateField",
Expand Down Expand Up @@ -294,7 +304,7 @@
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"targets": [
{
"datasource": {
Expand Down Expand Up @@ -381,7 +391,7 @@
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"targets": [
{
"datasource": {
Expand Down Expand Up @@ -527,7 +537,7 @@
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"targets": [
{
"datasource": {
Expand Down Expand Up @@ -708,7 +718,7 @@
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"targets": [
{
"datasource": {
Expand Down Expand Up @@ -779,7 +789,7 @@
"content": "#### VBIOS\n${g_gpu_vbios}\n\n#### DRIVER\n${g_driver}",
"mode": "markdown"
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"title": "Software",
"type": "text"
},
Expand Down Expand Up @@ -833,7 +843,7 @@
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"targets": [
{
"datasource": {
Expand Down Expand Up @@ -907,7 +917,7 @@
"textMode": "auto",
"wideLayout": true
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"targets": [
{
"datasource": {
Expand Down Expand Up @@ -984,7 +994,7 @@
"textMode": "value",
"wideLayout": true
},
"pluginVersion": "11.2.2+security-01",
"pluginVersion": "11.2.2",
"targets": [
{
"datasource": {
Expand Down Expand Up @@ -1348,17 +1358,30 @@
},
"disableTextWrap": false,
"editorMode": "builder",
"expr": "gpu_package_power{gpu_uuid=\"$g_gpu_uuid\"}",
"expr": "gpu_package_power{gpu_uuid=\"$g_gpu_uuid\", card_model=~\"102-G30211-00|102-G30211-0C|102-G30211-4C|102-G30212-0C|102-G30213-00|102-G30213-0C\"}",
"fullMetaSearch": false,
"includeNullMetadata": true,
"instant": false,
"legendFormat": "__auto",
"range": true,
"refId": "A",
"useBackend": false
},
{
"datasource": {
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"editorMode": "builder",
"expr": "gpu_average_package_power{gpu_uuid=\"$g_gpu_uuid\", card_model!~\"102-G30211-00|102-G30211-0C|102-G30211-4C|102-G30212-0C|102-G30213-00|102-G30213-0C\"}",
"hide": false,
"instant": false,
"legendFormat": "__auto",
"range": true,
"refId": "B"
}
],
"title": "GPU Power",
"title": "GPU Package Power",
"type": "timeseries"
},
{
Expand Down Expand Up @@ -1452,7 +1475,7 @@
},
"disableTextWrap": false,
"editorMode": "builder",
"expr": "gpu_edge_temperature{gpu_uuid=\"$g_gpu_uuid\", card_model!=\"$g_mi300_model\"}",
"expr": "gpu_edge_temperature{gpu_uuid=\"$g_gpu_uuid\", card_model!~\"102-G30211-00|102-G30211-0C|102-G30211-4C|102-G30212-0C|102-G30213-00|102-G30213-0C\"}",
"fullMetaSearch": false,
"includeNullMetadata": true,
"instant": false,
Expand All @@ -1468,7 +1491,7 @@
},
"disableTextWrap": false,
"editorMode": "builder",
"expr": "gpu_junction_temperature{gpu_uuid=\"$g_gpu_uuid\", card_model=\"$g_mi300_model\"}",
"expr": "gpu_junction_temperature{gpu_uuid=\"$g_gpu_uuid\", card_model=~\"102-G30211-00|102-G30211-0C|102-G30211-4C|102-G30212-0C|102-G30213-00|102-G30213-0C\"}",
"fullMetaSearch": false,
"hide": false,
"includeNullMetadata": true,
Expand Down Expand Up @@ -1735,16 +1758,16 @@
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"definition": "label_values(gpu_uuid)",
"hide": 0,
"definition": "label_values({hostname=\"$g_hostname\", gpu_id=\"$g_gpu_id\"}, gpu_uuid)",
"hide": 2,
"includeAll": false,
"label": "GPU UUID",
"multi": false,
"name": "g_gpu_uuid",
"options": [],
"query": {
"qryType": 1,
"query": "label_values(gpu_uuid)",
"query": "label_values({hostname=\"$g_hostname\", gpu_id=\"$g_gpu_id\"}, gpu_uuid)",
"refId": "PrometheusVariableQueryEditor-VariableQuery"
},
"refresh": 2,
Expand All @@ -1759,15 +1782,16 @@
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"definition": "label_values({gpu_uuid=\"$g_gpu_uuid\"},card_series)",
"hide": 2,
"definition": "label_values(hostname)",
"hide": 0,
"includeAll": false,
"label": "Hostname",
"multi": false,
"name": "g_card_series",
"name": "g_hostname",
"options": [],
"query": {
"qryType": 1,
"query": "label_values({gpu_uuid=\"$g_gpu_uuid\"},card_series)",
"query": "label_values(hostname)",
"refId": "PrometheusVariableQueryEditor-VariableQuery"
},
"refresh": 2,
Expand All @@ -1782,15 +1806,16 @@
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"definition": "label_values({gpu_uuid=\"$g_gpu_uuid\"},gpu_id)",
"hide": 2,
"definition": "label_values({hostname=\"$g_hostname\"},gpu_id)",
"hide": 0,
"includeAll": false,
"label": "GPU ID",
"multi": false,
"name": "g_gpu_id",
"options": [],
"query": {
"qryType": 1,
"query": "label_values({gpu_uuid=\"$g_gpu_uuid\"},gpu_id)",
"query": "label_values({hostname=\"$g_hostname\"},gpu_id)",
"refId": "PrometheusVariableQueryEditor-VariableQuery"
},
"refresh": 2,
Expand Down Expand Up @@ -1874,18 +1899,18 @@
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"definition": "label_values({gpu_uuid=\"$g_gpu_uuid\"},card_model)",
"definition": "label_values({gpu_uuid=\"$g_gpu_uuid\"},card_series)",
"hide": 2,
"includeAll": false,
"multi": false,
"name": "g_card_model",
"name": "g_card_series",
"options": [],
"query": {
"qryType": 1,
"query": "label_values({gpu_uuid=\"$g_gpu_uuid\"},card_model)",
"query": "label_values({gpu_uuid=\"$g_gpu_uuid\"},card_series)",
"refId": "PrometheusVariableQueryEditor-VariableQuery"
},
"refresh": 1,
"refresh": 2,
"regex": "",
"skipUrlSync": false,
"sort": 0,
Expand All @@ -1897,15 +1922,15 @@
"type": "prometheus",
"uid": "${DS_PROMETHEUS}"
},
"definition": "label_values({gpu_uuid=\"$g_gpu_uuid\"},hostname)",
"definition": "label_values({gpu_uuid=\"$g_gpu_uuid\"},card_model)",
"hide": 2,
"includeAll": false,
"multi": false,
"name": "g_hostname",
"name": "g_card_model",
"options": [],
"query": {
"qryType": 1,
"query": "label_values({gpu_uuid=\"$g_gpu_uuid\"},hostname)",
"query": "label_values({gpu_uuid=\"$g_gpu_uuid\"},card_model)",
"refId": "PrometheusVariableQueryEditor-VariableQuery"
},
"refresh": 2,
Expand All @@ -1931,31 +1956,11 @@
"query": "label_values({gpu_uuid=\"$g_gpu_uuid\"},serial_number)",
"refId": "PrometheusVariableQueryEditor-VariableQuery"
},
"refresh": 1,
"refresh": 2,
"regex": "",
"skipUrlSync": false,
"sort": 0,
"type": "query"
},
{
"hide": 2,
"label": "MI300 Model",
"name": "g_mi300_model",
"query": "${VAR_G_MI300_MODEL}",
"skipUrlSync": false,
"type": "constant",
"current": {
"value": "${VAR_G_MI300_MODEL}",
"text": "${VAR_G_MI300_MODEL}",
"selected": false
},
"options": [
{
"value": "${VAR_G_MI300_MODEL}",
"text": "${VAR_G_MI300_MODEL}",
"selected": false
}
]
}
]
},
Expand Down
Loading