Re: [PATCH v2 2/3] net/mlx5: add new devarg for Tx queue consecutive memory

[Thomas Monjalon <thomas@monjalon.net>]

26/06/2025 16:29, Stephen Hemminger:
> On Thu, 26 Jun 2025 13:18:18 +0000
> Bing Zhao <bingz@nvidia.com> wrote:
> 
> > Hi Stephen,
> > 
> > Thanks for your review and comments. I will add the description about the new devarg in our mlx5.rst file to have a detailed description.
> > Indeed, after some review and internal call discussion with our datapath experts. We would like to change the devarg a little bit but not only 0 / 1 as a chicken bit.
> > 
> > Since the memory accessing footprints and orders may impact the performance. In the perf test, we found that the alignment of the queue address may impact it. The basic starting address alignment is system page size, but it can be bigger.
> > So the new devarg use will be the log value of the alignment for all queues’ starting addresses. And on different CPU architectures / generations that have different LLC systems can try to use different alignment to get the best performance without rebuilding the binary application from the source code and it is configurable. WDYT?
> 
> Please make it automatic, the driver already has too many config bits.
> The users should just get good performance with the default.
> If driver needs to it should look at any/all system info to determine what the best setting is.

The default would be good enough, but it looks very difficult to know the best value.
That's why the idea is to allow some tuning here.

I agree there are too many devargs for most users,
but at the same time DPDK is a project to get the best performance,
that's why there are so many tuning possible.
I think we can meet both goals by making clear which options are for advanced tuning,
so a basic user will not mess with with such options which have a good default by the way.