CPC G06F 13/26 (2013.01) [G06F 3/0659 (2013.01); G06F 3/0679 (2013.01); G06F 12/0638 (2013.01); G06F 12/0866 (2013.01); G06F 13/14 (2013.01); G06F 13/1668 (2013.01); G06F 13/1684 (2013.01); G06F 13/28 (2013.01); G06F 13/364 (2013.01); G06F 13/4027 (2013.01); G06F 13/4068 (2013.01); G06F 13/4282 (2013.01); G06F 21/76 (2013.01); G06F 2213/0062 (2013.01); G06F 2213/40 (2013.01); G06F 2221/2103 (2013.01)] | 17 Claims |
1. A system, comprising:
a first die with a central processing unit (CPU) and a first bridge;
a second die with a second bridge, wherein the second die comprises a second CPU; and
die-to-die interconnects electrically coupled to the first bridge and the second bridge, wherein the CPU and the second CPU are configured to jointly arbitrate shared use of a resource in the system, wherein the resource comprises at least a subset of the die-to-die interconnects that are selectively used for bidirectional communication from the first die to the second die or from the second die to the first die and a given communication using at least the subset of die-to-die interconnections is unidirectional, wherein, in response to contention among a group of sources associated with services requests to access at least the subset of the die-to-die interconnects, the joint arbitration prioritizes one of the group of sources using one of: round-robin servicing, or guaranteed maximum latency scheduling; and
wherein, at a given time, a subset of signal lines in the die-to-die interconnects are allocated to the CPU and a second subset of the signal lines in the die-to-die interconnects are allocated to the second CPU.
|