Re: [dpdk-dev] [PATCH v8 1/3] eal/x86: run-time dispatch over memcpy

["Li, Xiaoyun" <xiaoyun.li@intel.com>]

> -----Original Message-----
> From: Thomas Monjalon [mailto:thomas@monjalon.net]
> Sent: Wednesday, October 25, 2017 17:00
> To: Li, Xiaoyun <xiaoyun.li@intel.com>
> Cc: Ananyev, Konstantin <konstantin.ananyev@intel.com>; Richardson,
> Bruce <bruce.richardson@intel.com>; dev@dpdk.org; Lu, Wenzhuo
> <wenzhuo.lu@intel.com>; Zhang, Helin <helin.zhang@intel.com>;
> ophirmu@mellanox.com
> Subject: Re: [dpdk-dev] [PATCH v8 1/3] eal/x86: run-time dispatch over
> memcpy
> 
> 25/10/2017 10:54, Li, Xiaoyun:
> > > > > > > I think we should focus on micro-benchmark and find a
> > > > > > > reasonnable threshold for a reasonnable drop tradeoff.
> > > > > > >
> > > > > > Sadly, it may not be that simple. What shows best performance
> > > > > > for
> > > > > > micro- benchmarks may not show the same effect in a real
> application.
> > > > > >
> > > > > > /Bruce
> > > > >
> > > > > Then how to measure the performance?
> > > > >
> > > > > And I cannot reproduce 15% drop on mellanox.
> > > > > Could the person who tested 15% drop help to do test again with
> > > > > 1024 threshold and see if there is any improvement?
> > > >
> > > > As Bruce said, best performance on micro-benchmark may not show
> > > > the
> > > same effect in real applications.
> > > > And I cannot reproduce the 15% drop.
> > > > And I don't know if raising the threshold can improve the perf or not.
> > > > Could the person who tested 15% drop help to do test again with
> > > > 1024
> > > threshold and see if there is any improvement?
> > >
> > > As I already asked before - why not to make that threshold dynamic?
> > > Konstantin
> > >
> > I want to confirm that raising threshold is useful. Then can make it dynamic
> and set it very large as default.
> 
> You can confirm it with micro-benchmarks.

I did tests on memcpy_perf_test. Set threshold to 1024.
But when smaller than 1024 bytes, it costs 2~4 cycles more than the original.
Such as original is 10, right now is 12. Then the drop is 2/12=16%.
I don't know this kind of drop matters a lot or not.
And above 1024 bytes, the drop is almost 4% as I said before.

/Xiaoyun