Offloading of computation for rack level servers and corresponding methods and systems

摘要

A method is disclosed that includes writing data to predetermined physical addresses of a system memory, the data including metadata that identifies a processing type; configuring a processor module to include the predetermined physical addresses, the processor module being physically connected to the memory bus by a memory module connection; and processing the write data according to the processing type with an offload processor mounted on the processor module.

主权项

1. A method, comprising:
issuing a write command on a system memory bus by operation of a memory controller, the system memory bus including address and data lines; receiving the write command from the system memory bus in an in-line module through an in-line module connector, an address of the write command identifying a request to at least one offload processor disposed on the in-line module; processing at least a portion of data included with the write command with the at least one offload processor to generate processed data; and transmitting the processed data from the in-line module over the system memory bus; wherein the system memory bus is further connected to at least one processor connector configured to receive at least one host processor different from the at least one offload processor, that can access a system memory via the system memory bus, and transmitting the processed data from the in-line module over the system memory bus includes transmitting the processed data to an input/output device that is different than a host processor.