[LITMUS^RT] Question about the scheduling overhead of GSN-EDF scheduler in LITMUS

[Björn Brandenburg]

> On 11 Oct 2015, at 16:57, Meng Xu <xumengpanda at gmail.com> wrote:
> 
> 
> ​I actually have one question about this: 
> Why the background cache-intensive task will affect the scheduling overhead? 
> Is it because the ready_queue data may be polluted by the background tasks so that it will take longer time to take one task from ready_queue and therefore increases the critical section? 
> I'm not sure if there is any other reason why the background tasks can increase the scheduling overhead. ​

Hi Meng, 

by running a cache-thrashing background workload, the cost of a memory accesses increases significantly. Basically, the rate of cache misses increases (scheduler data structures are more likely to be evicted between invocations of the scheduler) and any cache miss becomes much more expensive (because the memory controller is under heavy load). Basically, to approximate near-worst-case performance, you need to create lots of potential for interference at all shared hardware resources.

- Björn