[dpdk-dev] [PATCH v1 3/3] ethdev: enhance the API for getting burst mode information

Damjan Marion (damarion) damarion at cisco.com
Sat Nov 2 20:21:10 CET 2019



> On 2 Nov 2019, at 09:38, Slava Ovsiienko <viacheslavo at mellanox.com> wrote:
> 
> Hi
>> -----Original Message-----
>> From: Liu, Yu Y <yu.y.liu at intel.com>
>> Sent: Saturday, November 2, 2019 8:56
>> To: Wang, Haiyue <haiyue.wang at intel.com>; Thomas Monjalon
>> <thomas at monjalon.net>
>> Cc: dev at dpdk.org; arybchenko at solarflare.com; Yigit, Ferruh
>> <ferruh.yigit at intel.com>; jerinjacobk at gmail.com; Ye, Xiaolong
>> <xiaolong.ye at intel.com>; Kinsella, Ray <ray.kinsella at intel.com>; Sun,
>> Chenmin <chenmin.sun at intel.com>; Slava Ovsiienko
>> <viacheslavo at mellanox.com>; Damjan Marion (damarion)
>> <damarion at cisco.com>; Liu, Yu Y <yu.y.liu at intel.com>
>> Subject: RE: [PATCH v1 3/3] ethdev: enhance the API for getting burst mode
>> information
>> 
>> Add Damjan from FD.io for awareness...
>> 
>> Hi Thomas,
>> 
>> Long time no see. Sorry I use outlook which is not friendly to community
>> email.
>> 
>>> Anyway I will propose to replace this API in the next release.
>> Will your plan be affected by API/ABI stable plan?
>> BTW, if you propose new change in next release, it will make DPDK
>> consumer(FD.io) to change again.
>> So even if it is not affected to the API/ABI stable plan, do we still have time
>> to get a solution for everyone in DPDK 19.11 with your
>> contribution/acceleration?
>> 
>>> I suspect a real hidden issue in Intel CPUs that you try to mitigate.
>> Please be rest assured it is not the case.
>> This request is just from one FD.io project internal bug " tx/rx burst function
>> is shown as nil" reported by Chenmin.
> 
> Why just the presenting string with function name (possible with suffix) is not enough?
> I would like to see this API  (strings approach) in mlx5 either, dropping the entire feature
> does not look nice, as for me.
> 
> We could consider some requirements for the name suffices to distinguish whether
> function uses vector instructions and which ones if any.
> 
>> My understanding is DPDK behavior was taken as bug for someone in FD.io
>> project and potentially will mislead other DPDK consumer.
> 
> Why does FD.io code want to know which vector extension is used by burst routines?
> Is it going to share/preserve some resources (registers, etc.)? Is it robust ?
> Burst routines might not know whether vector extensions is used (they might call 
> libraries, even rte_memcpy() can use vectors in implicit fashion).

This is jut debug CLI print, it was added by me as a result of frustration of dealing constantly with
operational issues where people are reporting lower performance caused by DPDK deciding
for variety of reasons to switch from vector PMD to scalar one.

> 
>> Haiyue is working with Chenmin to address the issue and with your support it
>> will be even better.
>> 
>> Your support will be highly appreciated!
>> 
>> Thanks & Regards,
>> Yu Liu
>> 
>> -----Original Message-----
>> From: dev <dev-bounces at dpdk.org> On Behalf Of Wang, Haiyue
>> Sent: Saturday, November 2, 2019 1:30 PM
>> To: Thomas Monjalon <thomas at monjalon.net>
>> Cc: dev at dpdk.org; arybchenko at solarflare.com; Yigit, Ferruh
>> <ferruh.yigit at intel.com>; jerinjacobk at gmail.com; Ye, Xiaolong
>> <xiaolong.ye at intel.com>; Kinsella, Ray <ray.kinsella at intel.com>; Sun,
>> Chenmin <chenmin.sun at intel.com>; Slava Ovsiienko
>> <viacheslavo at mellanox.com>
>> Subject: Re: [dpdk-dev] [PATCH v1 3/3] ethdev: enhance the API for getting
>> burst mode information
>> 
>>> -----Original Message-----
>>> From: Thomas Monjalon <thomas at monjalon.net>
>>> Sent: Saturday, November 2, 2019 06:46
>>> To: Wang, Haiyue <haiyue.wang at intel.com>
>>> Cc: dev at dpdk.org; arybchenko at solarflare.com; Yigit, Ferruh
>>> <ferruh.yigit at intel.com>; jerinjacobk at gmail.com; Ye, Xiaolong
>>> <xiaolong.ye at intel.com>; Kinsella, Ray <ray.kinsella at intel.com>; Sun,
>>> Chenmin <chenmin.sun at intel.com>; Slava Ovsiienko
>>> <viacheslavo at mellanox.com>
>>> Subject: Re: [PATCH v1 3/3] ethdev: enhance the API for getting burst
>>> mode information
>>> 
>>> Thank you for trying to address comments done late.
>>> 
>>> 31/10/2019 18:11, Haiyue Wang:
>>>> --- a/lib/librte_ethdev/rte_ethdev.h
>>>> +++ b/lib/librte_ethdev/rte_ethdev.h
>> 
>> 
>>>> +#define RTE_ETH_BURST_ALTIVEC       (1ULL << 2)
>>>> +#define RTE_ETH_BURST_NEON          (1ULL << 3)
>>>> +#define RTE_ETH_BURST_SSE           (1ULL << 4)
>>>> +#define RTE_ETH_BURST_AVX2          (1ULL << 5)
>>>> +#define RTE_ETH_BURST_AVX512        (1ULL << 6)
>>> 
>>> Of course, I still believe that giving a special treatment to vector
>>> instructions is wrong.
>>> You did not justify why it needs to be defined in bits instead of
>>> string. I am not asking again because anyway you don't really reply. I
>>> think you are executing an order you received and I don't want to
>>> blame you more.
>>> I suspect a real hidden issue in Intel CPUs that you try to mitigate.
>>> No need to reply to this comment.
>>> Anyway I will propose to replace this API in the next release.
>> 
>> Never mind, if this design is truly ugly, drop it all now. I also prefer to do the
>> best, that's why open source is amazing, thanks! ;-)
> 



More information about the dev mailing list