* Rx performance
@ 2025-07-16 12:29 Lombardo, Ed
2025-07-16 12:45 ` Ivan Malov
0 siblings, 1 reply; 2+ messages in thread
From: Lombardo, Ed @ 2025-07-16 12:29 UTC (permalink / raw)
To: users
[-- Attachment #1: Type: text/plain, Size: 431 bytes --]
Hi,
I have E810-C 2x100G NIC, the perf tool shows scalar rx path and not vectorized rx path. When I turn off Rx Offload (NIC rx timestamp) I see vectorized rx path in perf tool.
Why is this a drawback and is this with just the E810 NIC?
is there a way around this so Rx Offload Rx Timestamp can be used in our application and get vectorized rx path?
We need rx timestamps of packets and max performance.
Thanks,
Ed
[-- Attachment #2: Type: text/html, Size: 730 bytes --]
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: Rx performance
2025-07-16 12:29 Rx performance Lombardo, Ed
@ 2025-07-16 12:45 ` Ivan Malov
0 siblings, 0 replies; 2+ messages in thread
From: Ivan Malov @ 2025-07-16 12:45 UTC (permalink / raw)
To: Lombardo, Ed; +Cc: users
[-- Attachment #1: Type: text/plain, Size: 704 bytes --]
Hi Ed,
On Wed, 16 Jul 2025, Lombardo, Ed wrote:
> Hi,
> I have E810-C 2x100G NIC, the perf tool shows scalar rx path and not vectorized rx path. When I turn off Rx Offload (NIC rx timestamp) I see vectorized rx path in perf tool.
>
> Why is this a drawback and is this with just the E810 NIC?
See the comment on line [1]. By the sound of it, this is somewhat expected.
[1] https://github.com/DPDK/dpdk/blob/fa3aca8a525decfd12bd5fa01ebc98789d5f7278/drivers/net/intel/common/rx.h#L104
Thank you.
>
> is there a way around this so Rx Offload Rx Timestamp can be used in our application and get vectorized rx path?
>
> We need rx timestamps of packets and max performance.
>
> Thanks,
> Ed
>
>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2025-07-16 12:45 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-07-16 12:29 Rx performance Lombardo, Ed
2025-07-16 12:45 ` Ivan Malov
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).