This is a combination of yangyicong and my notes. If we missed anything please point it out.
RMR: - (Shameer) Request to look at Patch 4 as some changes from earlier versions. Warning fix for next version. - Lorenzo will check with Robin. - Need to ask Joerg to pull the series. vCPU HP. - https://gitlab.com/jic23/qemu/-/commits/vcpu-poc-1 - Kernel Patches - hopefully posted soon. Not Lorenzo as not available for a few months. - _MAT needed or not. SPDM - Random discussion of the difficulty of debugging the rather complex exchanges. Nothing worth noting.
Scheduler: Yangyicong's slides provide a good summary of the question. - Observed that any topology based estimate is going to be challenging! - Vincent - set migration cost per sched domain level? - Dietmar - workloads weren't clear in original thread. Question of taskhot or new Idle Balance as relevant to workload. Tighter description of what is going on in the benchmarks needed. - Vincent - Uarch + cache etc relevant. - Hesham - potentially user perf counters to get some more info.
Takeways 1. Need to figure out the underlying reason of the performance variation of certain benchmark: If it's because of task hotness or newidle_balance(), etc. 2. Effect may related to the micro-arch, cache, and task's states. Also can be tuned according to the cpu numbers and scheduler domain levels. 3. Can be get from some firmware reports, to avoid the long time measurement in booting. 4. May be possible to calibrate during the boot time, but narrow the scope of measured CPUs and test time. This won't take long, but is susceptible to noise from other sources. 5. Hardware counters or profilings help understand what is going on, but unlikely to be consistently available for use in the loop. 6. Maybe make migration cost per sched domain on implementation.