On 4/26/2022 8:42 PM, Honnappa Nagarahalli wrote:
<snip>
Add support for using hugepages for worker lcore stack memory. The intent is
to improve performance by reducing stack memory related TLB misses and also
by using memory local to the NUMA node of each lcore.
This is a good idea. Have you measured any performance differences with this
patch? What kind of benefits do you see?
The performance benefit is very application and target dependent. For
example, a trivial test with a small memory/stack footprint would not
see much benefit from this change. An application with deep call stacks
and/or significant stack and table data memory usage would benefit.
We're not yet prepared to release specific performance delta data, but
we are currently developing applications that benefit from this feature.
All other comments will be incorporated in the next version.
