* Usage of rte_latencystats
@ 2024-09-27 10:53 Kamil Vojanec
0 siblings, 0 replies; only message in thread
From: Kamil Vojanec @ 2024-09-27 10:53 UTC (permalink / raw)
To: dev; +Cc: reshma.pattan, Jan Viktorin
[-- Attachment #1: Type: text/plain, Size: 1218 bytes --]
Hello,
I was experimenting with measuring the latency of DPDK applications
using the rte_latencystats module. While I could eventually get it to
work somehow, I noticed some issues, that led me to the following questions:
1. How to measure latency on multicore systems? I noticed that there are
a lot of global variables that store intermediate results of latency
calculations. Also, the documentation suggests calling
rte_latencystats_update() only on a single selected lcore.
2. What is a good value to select for the `samp_intvl` parameter of
rte_latencystats_init()? It seems that this option has a significant
impact on the measured application's performance.
3. How often is it recommended to call rte_latencystats_update()? The
documentation says the function needs to be called periodically, but
never specifies how often. Should this be done in each processing loop?
4. Is there no way to clear the statistics? I noticed that the minimum
and maximum measurement typically reaches some kind of static value that
does not change and does not really reflect the dynamic nature of the
measured system.
Looking forward to your reply,
Kamil Vojanec
CESNET, z.s.p.o.
[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4319 bytes --]
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2024-09-27 10:53 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-09-27 10:53 Usage of rte_latencystats Kamil Vojanec
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).