2020 ACRN Project Technical Community Meeting Minutes - WW10'20
Meeting Minutes Archive: Google doc: https://drive.google.com/drive/u/2/folders/111m6iU3PjSZebOOKzLagYPUWMvseGLeU
ACRN Project TCM - 4th Mar 2020
Attendees (Total 62, 4/3)
Note: If need to edit this document, please ask for access. We disabled anonymous editing to keep track of changes and identify who are the owners of the opens and agenda items.
Note: When adding opens or agenda items, please provide details (not only links), add your name next to the item you have added and specify your expectation from the TCM
2 months training session for ACRN new feature and architecture.
Download foil from ACRN Presentation->ACRN_TCM->WW10’20
Short description: Currently, IORR is supported in ACRN Hypervisor, it has good performance for I/O intensive task, but has the risk of I/O attack. Besides, it is unfair. So, BVT is introduced. It is a virtual time based scheduling algorithm, can provide fair share of pCPU to vCPU, and avoid I/O attack. In this session, we will talk about the implementation for BVT, and give a brief comparison for BVT and other scheduler algorithms.
Q1: GCC and bigsim are 2 examples of threads, 3 threads case is in Page 10.
Q2: Each thread of CPU workload is based on tick, no I/O and others.
Q3: MCU is Minimum charging Unit, physical number to 1 us.
Q4: Do not support CPU sharing for pre-launch VM in Hybrid mode yet, only for sharing VMs.
Q5: Could support latency sensitive VM, statically setup warp for latency sensitive thread.
Best & Regards