When a build trigger to the resource that have multiple agents.
Sometime QB give build to weaker agent.
I talked that because with same build, for example agent A took 2(weakest) hours to done job, agent B (stronger) took 1.5 hours to done job, agent C (strongest) took 1 hour.
Sometimes, the agent was given build by order: A -> C ->B.
I was confused how QB choose the agent to give the job.

I found that agent have a field benchmark. This field' purpose is to define the performance of agent, right?
If this is true, I saw this function benchmark() only calculate the single thread function. This can misleading to the wrong order of chosen agent.

@robinshen @Developers please help