System and method for cooling multiple logic modules
View Patent ↗A method and system of cooling hardware, sensing hardware, and supporting code streams that enables a non-redundant liquid cooling system to be used seemlessly in conjunction with an air cooling backup solution. The result offers system speed and reliability benefits of liquid cooling except for the brief occasions when the primary cooling system has failed and air cooling takes over. Until the liquid cooling is repaired, system clocks are automatically slowed to be compatible with the circuit speeds possible at the higher temperatures associated with air cooling. When the liquid cooling is repaired, the system clocks are automatically returned to their normal fast state. Depending on the circuit technology used, a supplied voltage may be varied while being air cooled.
1. A method to determine an optimum common clock speed for a server including multiple logic module each module being cooled by a cooling system, the method comprising:
determining a thermal state of each logic module of the multiple logic modules, each thermal state defined by a discrete temperature range; and
determining the optimum common clock speed based on module thermal states, self-test clock speeds, power state and cooling system repair status.
2. A method to determine a proper clock cycle time for multiple logic modules with different heat loads, the method comprising:
determining a thermal state of each logic module of the multiple logic modules, each thermal state defined by a discrete temperature range associated with a clock speed predetermined to be a proper clock cycle time for the temperature range;
determining whether a primary cooling means has been repaired; and
controlling a fan speed of the backup cooling fan to prevent oscillation between thermal states.
3. A method to determine a proper clock cycle time for multiple logic mondules with different heat loads, the method comprising:
determining a thermal state of each logic module of the multiple logic modules, each thermal state defined by a discrete temperature range associated with a clock speed predetermined to be a proper clock cycle time for the temperature range;
determining whether a primary cooling means has been repaired, and
increasing a voltage applied to a logic module to optimally use at least one of available cooling and power when operating in a backup cooling mode for maximum clock speed at a given temperature.
4. A method to determine a proper clock cycle time for multiule logic modules with different heat loads, the method comprising:
determining a thermal state of each logic module of the multiple logic modules, each thermal state defined by a discrete temperature range associated with a clock speed predetermined to be a proper clock cycle time for the temperature range;
determining whether a primary cooling means has been repaired;
decreasing a voltage applied to a logic module when operating in a backup cooling mode and at least one of cooling or power is unavailable to reduce leakage currents that warmer degraded temperatures generate.
5. A method to determine a proper clock cycle time for multiple logic modules with different heat loads, the method comprising:
determining a thermal state of each logic module of the multiple logic modules, each thermal state defined by a discrete temperature range associated with a clock speed predetermined to be a proper clock cycle time for the temperature range;
determining whether a primary cooling means has been repaired;
wherein said determining a thermal state of each logic module of the multiple logic modules is done through redundant thermal sensors directly monitoring a region representative of circuit temperatures of a corresponding logic module to provide at least one of thermal protection and redundancy to guide cooling control.
6. The method of claim 5 , wherein the region corresponds with one of a hat, substrate, and individual chips of a multi chip module (MCM).
7. The method of claim 5 , wherein the thermal sensors are compared for at least one of miscompare properties and insanity limits to check accuracy of each measured temperature.
8. The method of claim 7 , wherein the thermal sensors include a first thermal sensor sensed by the refrigerant unit and second and third thermal sensors read by a power supply supplying power to the multiple logic modules to insure at least one of full redundancy and accuracy.
9. The method of claim 8 , wherein the second and third thermal sensors are compared to each other and to the first thermal sensor, the second and third thermal sensors providing thermal protection of the multiple logic modules by dropping power if at least one of second and third thermal sensors indicate a temperature corresponding to a damage limit.
10. A method to determine a proper clock cycle time for multiple logic modules with different heat loads, the method comprising:
determining a thermal state of each logic module of the multiple logic modules, each thermal state defined by a discrete temperature range associated with a clock speed predetermined to be a proper clock cycle time for the temperature range;
determinig whether a primary cooling means has been repaired;
operating the backup air cooling fans in a manner to insure that the fans always turn on even if the primary cooling means has failed; and
operating the backup air cooling fans in a manner to insure that the fans do not come on so soon as to cause an oscillation of a cooling state when the primary cooling means has failed.