Slot airflow based on a configuration of the chassis
An information handling system includes a chassis having multiples sleds and an embedded controller. The embedded controller retrieves relative impedances for all of the sleds, and calculates a maximum available airflow for the first sled based the relative impedances of all other sleds. A baseboard management controller (BMC) of a first sled requests a boot operation for the first sled. The BMC collects configuration information for the first sled, and determines an airflow impedance of the first sled based on the configuration information. The BMC provides the airflow impedance and a power allocation request to the embedded controller. The BMC compares the maximum available airflow to a minimum airflow requirement for the first sled. If the maximum available airflow is less than the minimum airflow requirement, the BMC implements power limits for processors in the first sled to prevent overheating of components within the first sled.
1. A chassis for an information handling system, the chassis comprising:
a plurality of sleds including a first sled; and
an embedded controller to communicate with each of the sleds, the embedded controller to:
retrieve relative impedances for all of the sleds; and
calculate, via a machine learning system, a maximum available airflow for the first sled based on an impedance of the sled and based on the relative impedances of all other sleds;
wherein the first sled includes a baseboard management controller, the baseboard management controller to:
request a boot operation for the first sled;
collect configuration information for the first sled;
determine an airflow impedance of the first sled based on the configuration information;
provide the airflow impedance and a power allocation request to the embedded controller;
compare the maximum available airflow to a minimum airflow requirement for the first sled; and
if the maximum available airflow is less than the minimum airflow requirement, then implement power limits for processors in the first sled to prevent overheating of components within the first sled.
2. The chassis of claim 1 , wherein the baseboard management controller further to provide a message indicating that the power limits are implemented based on the maximum available airflow being less than the minimum airflow requirement for the first sled.
3. The chassis of claim 1 , wherein the calculation of the maximum available airflow, the machine learning system of the embedded controller further to calculate an airflow equation for the maximum available airflow for the first sled, wherein the airflow equation is calculated in one or more hidden layers of the machine learning system.
4. The chassis of claim 3 , wherein the airflow equation is a linear regression equation.
5. The chassis of claim 3 , wherein the airflow equation includes different weights for relative impedances of all of the sleds in the chassis.
6. The chassis of claim 1 , wherein the baseboard management controller further to determine the minimum airflow requirement based on thermal design point for the processors in the first sled.
7. The chassis of claim 1 , wherein the embedded controller further to generate multiple design of experiments, wherein each of the design of experiments includes a different relative impedance for all of the sleds in the chassis, and to train the machine learning system based on the multiple design of experiments.
8. The chassis of claim 7 , wherein the configuration information includes a power rating for processor of the sled and a number of hard disk drives in the first sled.
9. A method comprising:
collecting, by a baseboard management controller of a sled, configuration information for the sled, wherein the sled is one of a plurality of sleds within a chassis of an information handling system;
determining an airflow impedance of the sled based on the configuration information;
providing the airflow impedance to an embedded controller of the chassis;
retrieving, by the embedded controller, relative impedances for all of the sleds in the chassis;
calculating, by a machine learning system, a maximum available airflow for the sled based on the relative impedance of the sled and based on the impedances of all other sleds in the chassis;
comparing, by the baseboard management controller, the maximum available airflow to a minimum airflow requirement for the sled; and
if the maximum available airflow is less than the minimum airflow requirement, then implementing power limits for processors in the sled to prevent overheating of components within the sled.
10. The method of claim 9 , further comprising:
providing a message indicating that the power limits are implemented based on the maximum available airflow being less than the minimum airflow requirement for the sled.
11. The method of claim 9 , wherein the calculating of the maximum available airflow further comprises:
calculating, by the machine learning system, an airflow equation for the maximum available airflow for the sled, wherein the airflow equation is calculated in one or more hidden layers of the machine learning system.
12. The method of claim 11 , wherein the airflow equation is a linear regression equation.
13. The method of claim 11 , wherein the airflow equation includes different weights for relative impedances of all of the sleds in the chassis.
14. The method of claim 9 , further comprising:
determining the minimum airflow requirement based on thermal design point for the processors in the sled.
15. The method of claim 9 , further comprising:
generating multiple design of experiments, wherein each of the design of experiments includes a different relative impedance for all of the sleds in the chassis; and
training the machine learning system based on the multiple design of experiments.
16. The method of claim 9 , wherein the configuration information includes a power rating for processor of the sled and a number of hard disk drives in the sled.
17. A method comprising:
requesting, by a baseboard management controller of a sled, a boot operation for the sled, wherein the sled is one of a plurality of sleds within a chassis of an information handling system;
collecting configuration information for the sled;
determining an airflow impedance of the sled based on the configuration information;
providing the airflow impedance and a power allocation request to an embedded controller of the chassis;
retrieving, by the embedded controller, relative impedances for all of the sleds in the chassis;
calculating, by a machine learning system, a maximum available airflow for the sled based on the relative impedance of the sled and based on the impedances of all other sleds in the chassis;
comparing, by the baseboard management controller, the maximum available airflow to a minimum airflow requirement for the sled; and
if the maximum available airflow is less than the minimum airflow requirement, then implementing power limits for processors in the sled to prevent overheating of components within the sled.
18. The method of claim 17 , wherein the calculating of the maximum available airflow further comprises:
calculating, by the machine learning system, an airflow equation for the maximum available airflow for the sled, wherein the airflow equation is calculated in one or more hidden layers of the machine learning system.
19. The method of claim 17 , further comprising:
generating multiple design of experiments, wherein each of the design of experiments includes a different relative impedance for all of the sleds in the chassis; and
training the machine learning system based on the multiple design of experiments.
20. The method of claim 17 , further comprising:
determining the minimum airflow requirement based on thermal design point for the processors in the sled.