[openib-general] Re: user-mode verbs on Itanium
Grant Grundler
iod00d at hp.com
Tue May 10 08:24:28 PDT 2005
On Sun, May 08, 2005 at 05:23:45PM +0300, Michael S. Tsirkin wrote:
> Your patch initialized tstamp[0] to get_clock which still isnt right
> since its take before post_send and not after that.
Yes, but I think I want the first measurement *before* post_send
since it seems to measure something different that's part of the
code path too.
I suspect it's measuring the first "TLB" miss of HCA.
I found it very interesting that the first send (or two)
was very expensive.
grant
More information about the general
mailing list