Multiple virtual machine system with efficient cache memory design

A computer system comprising:

• a processor for issuing memory read and writes for a current partition of a plurality of partitions;
• a current partition cache to receive the current partition memory reads and writes and for resolving a memory read hit and a memory miss;
• a next partition cache selected when the next partition becomes active on the processor to receive next partition memory reads and writes from the processor;
• and
• an external memory for providing data to the processor when the current partition cache resolves the memory miss, and for providing appropriate data to the next partition cache to restore the next partition cache to a previous state.

A computer system comprising:

• a microprocessor for issuing memory reads and writes for a current virtual machine of a plurality of virtual machines operating on the microprocessor;
• a partition management unit for selecting the current virtual machine and for resolving memory read and write accesses;
• a current partition cache selected by the partition management unit for the current virtual machine to receive the current virtual machine memory reads and writes from the microprocessor and for resolving a memory read hit and a memory miss;
• a next partition cache selected by the partition management unit when the next virtual machine becomes active on the microprocessor to receive next partition memory reads and writes from the microprocessor;
• and
• an external memory for providing data to the microprocessor when the current partition cache resolves the memory miss and for providing appropriate data to the next partition cache to restore the next partition cache to a previous state.

A method of caching in a computer system in which a plurality of virtual machines are established, comprising the steps of:

• executing instructions in a microprocessor for a current partition of a plurality of partitions;
• issuing memory read and write instructions from the microprocessor to a partition management unit for the current partition;
• directing valid memory reads and writes to a current partition cache by the partition management unit;
• resolving memory reads in the current partition cache;
• resolving memory misses by reading an external memory to the current partition cache;
• reading appropriate frame blocks from the external memory to a next partition cache to restore the next partition cache to a previous state of a next partition to be active;
• and
• switching the current partition cache to the next partition cache when the next partition becomes active after a gap in time in switching from the current partition to the next partition.

A method of caching in a computer system in which a plurality of virtual machines are established, comprising the steps of:

• executing instructions in a microprocessor for a current partition of a plurality of partitions;
• issuing memory read and write instructions from the microprocessor to a current partition cache;
• resolving memory reads in the current partition cache;
• resolving memory misses by reading an external memory to the current partition cache;
• reading appropriate data from the external memory to a next partition cache to restore the next partition cache to a previous state of a next partition to be active;
• and
• switching the current partition cache to the next partition cache when the next partition becomes active after a gap of time.

A method of cache memory switching in a multiple process microprocessor system wherein an operating system of a microprocessor performs the steps comprising:

• determining which process is going to be an active next process;
• storing a current cache and a current microprocessor state for a current process to a memory during a context switch;
• and
• restoring a next cache and a next microprocessor state for the next process from the memory during the context switch.