Abstract: A method of regulating power for multi-node computer system components has a closed-ring path that links all the power governors and circulating in the ring is a system power number that represents the power consumption of the entire system. Meanwhile, all the governors keep counting its local power consumption. Each time the number passes a governor, the governor will add its local count onto this number, store this number for future usage, and reset its local count. When the new number returns back to the same power governor, the governor will subtract the new number with its stored number to calculate the overall system power usage within a number circulation period. The system power number overflow problem is also detected with a counter if the incoming number is smaller then the number previously stored. The counter whose counting capacity is greater than the maximum system power usage on all the nodes within a number circulation period.

Abstract: A method for regulating system power using a power governor for DRAM in a multi-node computer system regulating memory power consumption of an entire computer system employs a closed ring that connects all the power governors within the system to enable them to work in concert so that each of the power governors has the knowledge of memory activities within the entire system. They then control and limit the memory usage based on a true overall measurement instead of just local measurement. Each nodal power governor has memory command counter, ring number receiver, ring number transmitter, governor activation controller, and memory traffic controller. Each nodal power governor counts the weight of memory command. The degree of limiting actual memory activities can be programmed when the governor is active. Besides, the command priorities can be adjusted in activation too. A hybrid ring structure can be employed with a nodal power structure to achieve the fastest number circulation speed economically.

Abstract: A packetized cascade memory system including a plurality of memory assemblies, a memory bus including multiple segments, a bus repeater module and a segment level sparing module. The bus repeater module is in communication with two or more of the memory assemblies via the memory bus. The segment level sparing module provides segment level sparing for the communication bus upon segment failure.

Abstract: A method of regulating power for multi-node computer system components has a closed-ring path that links all the power governors and circulating in the ring is a system power number that represents the power consumption of the entire system. Meanwhile, all the governors keep counting its local power consumption. Each time the number passes a governor, the governor will add its local count onto this number, store this number for future usage, and reset its local count. When the new number returns back to the same power governor, the governor will subtract the new number with its stored number to calculate the overall system power usage within a number circulation period. The system power number overflow problem is also detected with a counter if the incoming number is smaller then the number previously stored. The counter whose counting capacity is greater than the maximum system power usage on all the nodes within a number circulation period.

Abstract: A power governor for DRAM in a multi-node computer system regulating memory power consumption of an entire computer system employs a closed ring that connects all the power governors within the system to enable them to work in concert so that each of the power governors has the knowledge of memory activities within the entire system. They then control and limit the memory usage based on a true overall measurement instead of just local measurement. Each nodal power governor has memory command counter, ring number receiver, ring number transmitter, governor activation controller, and memory traffic controller. Each nodal power governor counts the weight of memory command. The degree of limiting actual memory activities can be programmed when the governor is active. Besides, the command priorities can be adjusted in activation too. A hybrid ring structure can be employed with a nodal power structure to achieve the fastest number circulation speed economically.

Abstract: A memory subsystem with positional read data latency that includes a cascaded interconnect system with one or more memory modules, a memory controller and one or more memory busses. The memory controller includes instructions for providing positional read data latency. The memory modules and the memory controller are interconnected by a packetized multi-transfer interface via the memory busses.

Abstract: A memory subsystem that includes segment level sparing. The memory subsystem includes a cascaded interconnect system with segment level sparing. The cascaded interconnect system includes two or more memory assemblies and a memory bus. The memory bus includes multiple segments and the memory assemblies are interconnected via the memory bus.

Abstract: A memory subsystem for providing a bus speed multiplier. The memory subsystem includes one or more memory modules operating at a memory module data rate. The memory subsystem also includes a memory controller and one or more memory busses. The memory busses operate at four times the memory module data rate. The memory controller and the memory modules are interconnected by a packetized multi-transfer interface via the memory busses.

Abstract: A packetized cascade memory system including a plurality of memory assemblies, a memory bus including multiple segments, a bus repeater module and a segment level sparing module. The bus repeater module is in communication with two or more of the memory assemblies via the memory bus. The segment level sparing module provides segment level sparing for the communication bus upon segment failure.

Abstract: Memory is scrubbed by an improved non-linear method giving scrubbing preference to the central storage region having the characteristic of a high risk read-only memory such as the CPA region to prevent the accumulation of temporary data errors. The chip row on which the CPA resides is scrubbed after each time the scrubbing of a non-CPA chip row in a PMA completed successfully. The next non-CPA least recently scrubbed chip row would be selected for scrubbing after scrubbing completed on the CPA chip row. This in a first case provides non-linear selection methods of scrubbing central storage of computer systems to more frequently select (“select” herein encompasses the meaning of “favor”) scrub regions having the characteristic of a predominately read-only memory making those regions at a higher risk of failure than those regions having lower risk because of frequent write operations.