SMART Embedded Computing MC1600, Инструкция по установке и эксплуатации

Страница: 116 / 126

116 MC1600 Extreme Edge Server Installation and Use (6806870A02B)
Software Configuration
Software Configuration
5.2.2.1.4 Aggregated Temperature Sensors
The aggregated temperature sensor is a logical sensor the BMC firmware implements for
the PCIe slot representing the thermal status of the PCIe card in the corresponding slot.
This logical sensor is provided so the user may include the thermal status of the PCIe card
to be included in the aggregated sensor to be used by the system cooling algorithm. The
BMC considers the thermal status along with other thermal information collected within the
chassis, and determines the speed of cooling fans.
The
Set Aggregated Temperature Sensor
command is used to update the aggregated
temperature sensor of the PCIe card. If the aggregated temperature sensor is never
updated, it is considered as disabled, and will not be part of the cooling algorithm. However,
if an aggregated temperature sensor has been updated through the
Set Aggregated
Temperature Sensor
command, the application or service keeps updating the sensor at
least once every 3 seconds, or the sensor reading will become invalid. An invalid
aggregated temperature makes the chassis-wide aggregated temperature invalid and as a
result, all fans will run at full speed. To avoid this from happening, the application or service
shall use the
Set Aggregated Temperature Sensor
command to disable the sensor
gracefully before it exits or stops.
The aggregated thermal status is tabevaluated in a 0 to 63 scale, where 0 means no
thermal concerns and cooling fans may be turned off or running at their minimum speed,
56 means the cooling fans should be running at full speed to avoid the further temperature
rising, and 63 indicates permanent thermal damage may happen. Typically, a temperature
between the IPMI temperature sensor upper non-critical threshold and the upper critical
threshold is linearly mapped to 0 to 56, and a temperature between the upper critical
threshold and the upper non-recoverable threshold is linearly mapped to 56 to 63. All
temperatures below the upper non-critical threshold are mapped to 0, and all temperatures
over the upper non-recoverable are mapped to 63.
Following is a simple pseudo code example showing how to implement the PCIe card
thermal monitoring.
for t_sensor in all_readable_temperature_sensors_on_the_pcie_card
temperature=read(t_sensor)
if temperature < non-critical[t_sensor]
aggtemp[t_sensor] = 0
else if temperature < critical[t_sensor]
aggtemp[t_sensor] = (56 - 0) * (temperature - non-
critical[t_sensor]) / (critical[t_sensor] - non-critical[t_sensor])
else if temperature < non-recoverable[t_sensor]
aggtemp[t_sensor] = (63 - 56) * (temperature - critical[t_sensor])
/ (non-recoverable[t_sensor] - critical[t_sensor])
else
aggtemp[t_sensor] = 63