Martin,
Can the critical path itself change based on concurrent load if each of the 
paths making up the graph has different serialised characteristics? I'm 
postulating that the path having minimum service time at N=1 but higher 
contention may never appear to be the critical path at lower concurrency 
but it will manifest at higher concurrency. 

Shripad

On Wednesday, September 20, 2017 at 10:10:37 PM UTC+5:30, Martin Thompson 
wrote:
>
> I've a different query on this. As the response time depends on number of 
>> concurrent users, and if there are parallel paths in the flow being 
>> measured, you will need to measure scalability characteristics of each 
>> path. Any thoughts on this?
>>
>
> It is often the case that a number of parallel paths can exist making up a 
> graph. If you map out the graph and determine the critical path you can 
> then get the contention component for what cannot be parallel, you also can 
> then get and service time for queuing theory.
>
> Even on a single thread this manifests as IPC (Instructions Per Cycle) as 
> the CPU extras parallelism which is not constrained by data dependencies.
>

-- 
You received this message because you are subscribed to the Google Groups 
"mechanical-sympathy" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to