Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] RDMA pipeline
From: Gleb Natapov (glebn_at_[hidden])
Date: 2008-02-20 04:29:57

On Tue, Feb 19, 2008 at 10:40:46PM -0500, George Bosilca wrote:
> Actually, it restores the original behavior. The RDMA operations were
> pipelined before the r15247 commit, independent of the fact that they
> had mpool or not. We were actively using this behavior in the message
> logging framework to hide the cost of the local storage of the payload,
> and we were quite surprised when we realized that it disappeared.
I checked v1.2 with tcp BTL (I can't test mx or elan, but tcp also
support RDMA and has no mpool) and no matter what btl_tcp_max_rdma_size
I provide the whole buffer is sent in one rdma operation. And here is
explanation why this happens:
 1. If BTL is RDMA capable but does not provide mpool
 mca_pml_ob1_rdma_btls() assumes that memory is always registered. This
 function will always return non zero value for any buffer it is called
 with in our case.

 2. When mca_pml_ob1_send_request_start_btl() chooses what function to
 use for rendezvous send it checks if buffer is contiguous and if it is
 then it check if buffer is already registered by checking non zero value
 returned by mca_pml_ob1_rdma_btls() and for BTLs without mpool
 mca_pml_ob1_send_request_start_rdma() is always chosen.

 3. Receiver checks if local buffer is registered by calling
 mca_pml_ob1_rdma_btls() on it (see pml_ob1_recvreq.c:259):

  recvreq->req_rdma_cnt = mca_pml_ob1_rdma_btls(
          (unsigned char*) base,
 So recvreq->req_rdma_cnt is set to non zero value (if receive buffer is
 contiguous of cause).

 4. Receiver send PUT messages to a senders in
 mca_pml_ob1_recv_request_schedule_exclusive(). Here is the code snip
 from the function (see pml_ob1_recvreq.c:684):

       /* makes sure that we don't exceed BTL max rdma size
        * if memory is not pinned already */
       if(0 == recvreq->req_rdma_cnt &&
             bml_btl->btl_max_rdma_size != 0 &&
             size > bml_btl->btl_max_rdma_size)

           size = bml_btl->btl_max_rdma_size;
 Pay special attention to a comment. If recvreq->req_rdma_cnt is not
 zero btl_max_rdma_size is ignored and message is send by one big RDMA

So what I have shown here is that there was no pipeline for TCP btl in
v1.2 and that the code specifically written to behave this way.
If you still think that there is a difference in behaviour between v1.2
and the trunk can you explain what code path is executed in v1.2 for
your test case and how trunk behaves differently.

> If a BTL don't want to use pipeline for RDMA operations, it can set the
> RDMA fragment size to the max value, and this will automatically disable
> the pipeline. However, if the BTL support pipeline with the trunk version
> today it is not possible to activate it. Moreover, in the current version
> the parameters that define the BTL behavior are blatantly ignored, as the
> PML make high level assumption about what they want to do.
I am not defending current behaviour. If you want to change it we can
discuss exact semantics that you want to see. But before that I what to
make sure that trunk is indeed different from v1.2 in this regard like
you claim it to be. Can you provide me with a test case that works
differently in v1.2 and the trunk?