[PATCH 1/6] service: reduce statistics overhead for parallel services

[Mattias Rönnblom]

On 2022-10-03 10:06, David Marchand wrote:
> On Tue, Sep 6, 2022 at 6:17 PM Mattias Rönnblom
> <mattias.ronnblom at ericsson.com> wrote:
>>
>> Move the statistics from the service data structure to the per-lcore
>> struct. This eliminates contention for the counter cache lines, which
>> decreases the producer-side statistics overhead for services deployed
>> across many lcores.
>>
>> Prior to this patch, enabling statistics for a service with a
>> per-service function call latency of 1000 clock cycles deployed across
>> 16 cores on a Intel Xeon 6230N @ 2,3 GHz would incur a cost of ~10000
>> core clock cycles per service call. After this patch, the statistics
>> overhead is reduce to 22 clock cycles per call.
>>
>> Signed-off-by: Mattias Rönnblom <mattias.ronnblom at ericsson.com>
> 
> Re-reading the mail archive, I found that Morten acked the series
> (replying on patch 4).
> Mattias, such a series deserved a cover letter.
> 

Noted. Do you want a v2 with a cover letter?

> How does this series fit with other patches from Harry?
> https://protect2.fireeye.com/v1/url?k=31323334-501d5122-313273af-454445555731-9ac8e99afde9bc77&q=1&e=946b2bf5-b58c-4c02-b173-bba36b0b12f9&u=https%3A%2F%2Fpatchwork.dpdk.org%2Fproject%2Fdpdk%2Flist%2F%3Fseries%3D23959%26state%3D%2A
> 
> Thanks.
> 

The patchset I submitted is made against the above-linked patchset. 
Nothing in the patchset I submitted depends on those patches. I just 
assumed that Harry's patches would have been merged at the point my 
patchset was considered.

I can base a v2 against the current main, if you so prefer.