Aggregating remote interconnected resources like Memory, CPU, GPU to local Hypervisor

[ohault]

Using high-speed interconnect (PCIe, PCIe interconnect, n x 10GbE, IB, ...), a major feature would be to support the concept of aggregating remote interconnected resources like Memory, CPU, GPU to local Hypervisor.

There are existing projects like :

• https://giantvm.github.io/
• https://github.com/systems-nuts/FragVisor

The OpenVMM/OpenHCL would already be well architected to support this.