[openib-general] Re: IPOIB_NUM_WC question

[Roland Dreier]

    Shirley> Hello, Roland, I have a question about below code. Why
    Shirley> IPOIB_NUM_WC was set to 4 by default? If there is no
    Shirley> particular reason, I would like to submit a patch to make
    Shirley> this value tunable.

No particular reason other than tuning for my particular test system.

    Shirley> I found some interesting result on different drivers when
    Shirley> making this number tunable. In some case, netperf could
    Shirley> even get up to 30% throughout on ehca, but mthca only
    Shirley> gets around 2%. I can post a patch and test result if you
    Shirley> are interested.

It might be good to make it a tunable.  But maybe there's one best
number.  It's bad to have too many obscure knobs for people to fool
around with.

I wonder if there's a way to make the driver self-tuning by keeping
some simple statistics while polling the CQ.

    Shirley> Also I noticed that mthca sends IB_CQ_NEXT_COMP with
    Shirley> completion entry==0 pretty often, in my netperf test,
    Shirley> more than 30% time.

I can't quite understand this.  What do you mean by sending IB_CQ_NEXT_COMP?

 - R.