Monitoring VRM-induced memory errors
View Patent ↗A method and system for improving Field Replacement Unit (FRU) isolation in memory sub-systems by monitoring Voltage Regulator Module (VRM) induced memory errors. A comparator compares the output voltage coming from the VRM to memory. If the comparator detects a VRM output voltage transient that is outside a rated threshold, then a counter is increased by one. If the counter exceeds a count threshold, a VRM error is posted. If a memory failure occurs within a predetermined period of time, then the VRM error pinpoints the VRM output voltage transient as being the likely cause of the memory failure.
1. A voltage regulator module failure detector, wherein the voltage regulator module failure detector is a hardware device comprising:
a voltage comparator comprising:
a first input, the first input receiving an output front a voltage regulator module (VRM) that supplies regulated power to a computer component,
a second input, the second input receiving a threshold voltage, the threshold voltage being a boundary of an operational output voltage of the VRM, wherein the threshold voltage is a minimum operational output voltage of the VRM, and
an output, the output transmitting a result of a comparison of the output from the VRM with the threshold voltage, wherein the result is transmitted only when the output from the VRM is outside the threshold voltage;
a courtier coupled to the output of the voltage comparator, wherein the counter counts the number of times that the output from the VRM is outside the threshold voltage, and wherein the counter is cleared if a failure of the computer component has not occurred after a pre-determined period of time from when the VRM error was posted;
an event comparator coupled to the counter, the event comparator comparing a count in the counter with a pre-determined event threshold, wherein a VRM error is posted by the event comparator when the count in the counter exceeds the pre-determined event threshold; and
a combinational logic having a first input of the posted VRM error and a second input of the computer component failure, wherein in response to the posted VRM error and computer component failure both being input contemporaneously into the combinational logic, the combinational logic outputs a message instructing a replacement of the VRM.
2. The voltage regulator module failure detector of claim 1 , wherein the threshold voltage is a maximum operational output voltage of the VRM.
3. The voltage regulator module failure detector of claim 1 , wherein the threshold voltage is a pre-determined percentage of an operational output voltage of the VRM.
4. The voltage regulator module failure detector of claim 1 , wherein the counter is cleared at a pre-determined interval of time that is independent of when a VRM error occurred.
5. The voltage regulator module failure detector of claim 1 , wherein the computer component is a memory.
6. The voltage regulator module failure detector of claim 1 , wherein the computer component is a microprocessor.
7. A computer system comprising:
a hardware voltage regulator module failure detector that includes:
a voltage comparator having:
a first input, the first input receiving an output from a voltage regulator module (VRM) that supplies regulated power to a computer component,
a second input, the second input receiving a threshold voltage, the threshold voltage being a boundary of an operational output voltage of the VRM, wherein the threshold voltage is a pre-determined percentage of the operational output voltage of the VRM, and
an output, the output transmitting a result of a comparison of the output from the VRM with the threshold voltage, wherein the result is transmitted only when the output from the VRM is outside the threshold voltage;
a counter coupled to the output of the voltage comparator, wherein the counter counts the number of times that the output from the VRM is outside the threshold voltage, and wherein the counter is cleared if a failure of the computer component has not occurred after a pre-determined period of time from when the VRM error was posted;
an event comparator coupled to the counter, the event comparator comparing a count in the counter with a pre-determined event threshold, wherein a VRM error is posted by the event comparator when the count in the counter exceeds the pre-determined event threshold; and
a combinational logic having a first input of the posted VRM error and a second input of the computer component failure, wherein in response to the posted VRM error and computer component failure both being input contemporaneously into the combinational logic, the combinational logic outputs a message instructing a replacement of the VRM.
8. The computer system of claim 7 , wherein the threshold voltage is a maximum operational output voltage of the VRM.
9. The computer system of claim 7 , wherein the threshold voltage is a minimum operational output voltage of the VRM.