Isolated shared memory architecture (iSMA)
Techniques for a massively parallel and memory centric computing system. The system has a plurality of processing units operably coupled to each other through one or more communication channels. Each of the plurality of processing units has an ISMn interface device. Each of the plurality of ISMn interface devices is coupled to an ISMe endpoint connected to each of the processing units. The system has a plurality of DRAM or Flash memories configured in a disaggregated architecture and one or more switch nodes operably coupling the plurality of DRAM or Flash memories in the disaggregated architecture. The system has a plurality of high speed optical cables configured to communicate at a transmission rate of 100 G or greater to facilitate communication from any one of the plurality of processing units to any one of the plurality of DRAM or Flash memories.
1. A massively parallel and memory centric computing system, the system comprising:
a plurality of processing units operably coupled to each other through one or more communication channels, each of the plurality of processing units having an ISMn (Isolated Shared Memory network) interface device, each of the plurality of ISMn interface devices being coupled to an ISMe (Isolated Shared Memory endpoint) devices connected to each of the processing units;
a plurality of DRAM or Flash memories configured in a disaggregated architecture;
one or more switch nodes operably coupling the plurality of DRAM or Flash memories in the disaggregated architecture; and
a plurality of high speed optical cables configured to communicate at a transmission rate of 100 G or greater to facilitate communication from any one of the plurality of processing units to any one of the plurality of DRAM or Flash memories.
2. The system of claim 1 wherein each of the plurality of high speed optical cables having a length of 1 meter to about 10 kilometers.
3. The system of claim 1 wherein the transmission rate is 100 G PAM or other protocol.
4. The system of claim 1 wherein the plurality of processing units is a number from 1 through N, where N is an integer greater than or equal to thirty two.
5. The system of claim 1 wherein each of the processing units is either an ARM or an Intel based x86 processor.
6. The system of claim 1 wherein the system is configured to initiate a power on or system boot, each of the iSMn interface devices being configured to determine a capacity of any one or all of the plurality of DRAM or Flash memories.
7. The system of claim 1 wherein each of the iSMn interface devices is configured to communicate in a broadcast process among any other iSMn interface device, each of which is coupled to at least one of the plurality of DRAM or Flash memories; whereupon the broadcast process is provided to determine a capacity and a topology of any or all of the system including the plurality of DRAM or Flash memories or networking configuration.
8. The system of claim 7 wherein the topology comprises information selected from at least one of a number of connected processing units and identification information of the processing units to the iSMn devices.
9. The system of claim 8 wherein each of the iSMn devices is configured to initiate communication of the topology and capacity information to any one or all of the iSMe devices using a communication direction from iSMn devices to the iSMe devices.
10. The system of claim 9 wherein each of the iSMe devices is configured to thereafter communicate the topology and a collective capacity of a sum-total of the capacity to a particular processing unit during a PCIe enumeration process.
11. The system of claim 10 wherein the sum-total memory capacity information is transferred to a particular processing unit as an address range in a PCIe endpoint base address register.
12. The system of claim 11 wherein transferring of the sum-total memory capacity is provided using a base address register (BAR) characterized by allowing each of the processing units to have a common address view of the disaggregated memory.
13. The system of claim 12 wherein the BAR range reporting of the disaggregated memory is configured to provide a mapping of a physical address range of the disaggregated memory into a common virtual address range, thereby configured to provide caching of a virtual to physical address translation of the disaggregated memory provided by a translation look-aside buffer in the processing unit.
14. The system of claim 13 wherein the common address view of the disaggregated memory is configured as a common virtual address.
15. A massively parallel and memory centric computing system, the system comprising:
a plurality of processing units operably coupled to each other through one or more communication channels;
an ISMe (Isolated Shared Memory endpoint) device coupled to each of the processing units;
an ISMn (Isolated Shared Memory network) interface device coupled to each of the ISMe devices;
a plurality of DRAM or Flash memories configured in a disaggregated architecture and coupled to the plurality of iSMn interface devices;
one or more switch nodes operably coupling the plurality of DRAM or Flash memories in the disaggregated architecture; and
a plurality of high speed optical cables configured to communicate at a transmission rate of 100 G or greater to facilitate communication from any one of the plurality of processing units to any one of the plurality of DRAM or Flash memories.
16. The system of claim 15 wherein each of the iSMn interface devices is configured to communicate in a broadcast process among any other iSMn interface device, each of which is coupled to at least one of the plurality of DRAM or Flash memories; whereupon the broadcast process is provided to determine a capacity and a topology of any or all of the system including the plurality of DRAM or Flash memories or networking configuration.
17. The system of claim 16 wherein the topology comprises information selected from at least one of a number of connected processing units and identification information of the processing units to the iSMn devices; and wherein each of the iSMn devices is configured to initiate communication of the topology and capacity information to any one or all of the iSMe devices using a communication direction from iSMn devices to the iSMe devices.
18. The system of claim 17 wherein each of the iSMe devices is configured to thereafter communicate the topology and a collective capacity of a sum-total of the capacity to a particular processing unit during a PCIe enumeration process; and wherein the sum-total memory capacity information is transferred to a particular processing unit as an address range in a PCIe endpoint base address register.
19. The system of claim 18 wherein transferring of the sum-total memory capacity is provided using a base address register (BAR) characterized by allowing each of the processing units to have a common address view of the disaggregated memory; and wherein the BAR range reporting of the disaggregated memory is configured to provide a mapping of a physical address range of the disaggregated memory into a common virtual address range, thereby configured to provide caching of a virtual to physical address translation of the disaggregated memory provided by a translation look-aside buffer in the processing unit.
20. The system of claim 19 wherein the common address view of the disaggregated memory is configured as a common virtual address.