[libfabric-users] FI_REMOTE_CQ_DATA & RMA
QUENTIN BOYER
quentin.boyer at atos.net
Mon Sep 26 07:42:17 PDT 2022
I don't really see which is the parameter that controls this for
connection-less endpoints, the destination address ?
Thank you!
Quentin
On 26/09/2022 16:34, Zhang, Wei wrote:
> Caution: External email. Do not open attachments or click links, unless this email comes from a known sender and you know the content is safe.
>
>
> Hi,
>
> When you can fi_write, you need to specify which end point you are targeting to (alongside with memory address).
>
> So only the target endpoint will generate completion.
>
> Wei
>
> On 9/26/22, 6:23 AM, "Libfabric-users on behalf of QUENTIN BOYER via Libfabric-users" <libfabric-users-bounces at lists.openfabrics.org on behalf of libfabric-users at lists.openfabrics.org> wrote:
>
> CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you can confirm the sender and know the content is safe.
>
>
>
> Hi,
>
> There is an interaction I don't quite understand. If a provider does
> *not* have FI_MR_ENDPOINT the memory regions are bound to domains. But
> when the are target with a RMA write that provides cq data with
> FI_REMOTE_CQ_DATA where should the completion be generated ? because
> there may be one, zero, or many endpoints associated with this domain.
>
> Or is this flag only applicable with FI_MR_ENDPOINT ?
>
> Quentin Boyer
>
> _______________________________________________
> Libfabric-users mailing list
> Libfabric-users at lists.openfabrics.org
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.openfabrics.org%2Fmailman%2Flistinfo%2Flibfabric-users&data=05%7C01%7Cquentin.boyer%40atos.net%7C4ce84d78248942aac6b708da9fcc4bf6%7C33440fc6b7c7412cbb730e70b0198d5a%7C0%7C0%7C637997997039177853%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=tYDExzY8knKOpCcQzVo5uuo9nV7wwWg5bwHeetQYoW4%3D&reserved=0
>
More information about the Libfabric-users
mailing list