Closed ArneTR closed 3 months ago
@ribalba Can you give it a quick look? Should be all fine though. IPMI I could not test, since we have no machine atm.
Eco-CI Output: | Label | 🖥 avg. CPU utilization [%] | 🔋 Total Energy [Joules] | 🔌 avg. Power [Watts] | Duration [Seconds] |
---|---|---|---|---|---|
Total Run | 22.6087 | 1406.7 | 3.51675 | 407 | |
Measurement #1 | 22.8862 | 1406.7 | 3.51675 | 401 |
📈 Energy graph:
8.18 ┤ â•â”€â”€â•®
7.54 ┤ │ │
6.90 ┤ â•â•® │ â•°â•®
6.26 ┤ ││ â•â•® â•â•®â”‚ │
5.62 ┤ ││ â•â•¯â•°â•® │││ │
4.97 ┤ ││ â•â•® â•â•¯ │ │╰╯ │
4.33 ┤ â•â”€â”€â”€â•® â•â•® â•â”€â•® â•â•® â•â•® │╰──╯╰──╮ │ ╰─╯ │ â•â•® â•â•® â•â•® â•â•® â•â•® â•â•® â•â•® â•â•® â•â•® â•â•®
3.69 ┤ â•â”€â•¯ â•°â•® │╰──────╮ â•â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â•® â•â”€â”€â”€â”€â”€â”€â”€â”€â•¯ ╰─╯╰╮ â•â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â•¯â•°â”€â”€â”€â”€â”€â”€â”€â”€â”€â•¯ │ │ │ â•â•¯â•°â”€â”€â•®â•â”€â”€â”€â•®â”‚│â•â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â•®â•â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â•® â•â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â•®â•â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â•¯â•°â”€â”€â”€â”€â•¯â•°â”€â”€â”€â•¯â•°â”€â”€â”€â”€â•¯â•°â”€â”€â”€â•¯â•°â”€â”€â”€â”€â•¯â•°â”€â”€â”€â”€â”€â”€â”€â”€â”€â•¯â•°â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â•® â•â”€â•®â•â”€â•®â•â”€â”€â”€â•®â•â”€â•®â•â”€â•®â•â•®â•â”€â•® â•â”€â”€â•® â•â”€â”€â”€â•®â•â”€â•®â•â”€â•®â•â”€â•®â•â”€â•®â•â”€â•®â•â”€â”€â”€â”€â”€â”€â”€â•® â•â•® â•â•®â•â”€â•¯â•°
3.05 ┼╮ â•â•®â”‚ │ â•â•¯ ╰╮│ │ │ â•°â•® │ â•°â•®â•â•¯ │ │ ││ ││││ ││ â•°â•® │ ╰╯ │ â•â•¯ ╰╯ ││ ││ ╰╯ ││││ │â•â•¯ │â•â•¯ ╰╯ ││ ││ ││ ││ ││ │ ││ â•â•¯â”‚│
2.41 ┤│ â•â•¯â•°â•¯ │ │ ││ │ │ │ │ ╰╯ │ â•â•¯ ╰╯ ││││ ││ â•°â•® │ │ │ ││ ││ ││╰╯ ││ ╰╯ ╰╯ ╰╯ ││ ╰╯ ││ │ │╰─╮ │ ││
1.77 ┤╰─────╯ ╰──╯ ╰╯ ╰─╯ ╰─╯ ╰────╯ ╰╯╰╯ ╰╯ ╰────────────────────╯ ╰─╯ ╰╯ ╰╯ ╰╯ ╰╯ ╰╯ ╰╯ ╰─────────╯ ╰───────╯ ╰╯
Watts over time
Some reporters where using
-i 0.1
to sleep for 100 ms and some-i 100
to sleep for 100 msThis is now normalized.
The initial design decision was to to the least amount possible in the reporters in terms of conversion when collecting data. However we could not measure any difference in energy or time when using the non-conversion approach. So we opted for better consistency.