IB/hfi1: Add adaptive cacheless verbs copy

Browse source

The kernel memcpy is faster than a cacheless copy.  However,
if too much of the L3 cache is overwritten by one-time copies
then overall bandwidth suffers.  Implement an adaptive scheme
where full page copies are tracked and if the number of unique
entries are larger than a threshold, verbs will use a cacheless
copy.  Tracked entries are gradually cleaned, allowing memcpy to
resume once the larger copies have stopped.

Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Jubin John <jubin.john@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

...

This commit is contained in:

Dean Luick 2016-03-05 08:50:43 -08:00 • committed by Doug Ledford

parent 8fefef125e

commit 528ee9fbf0

3 changed files with 211 additions and 2 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download patch file Download diff file Expand all files Collapse all files

6

drivers/staging/rdma/hfi1/init.c

View file

185

drivers/staging/rdma/hfi1/verbs.c

View file

22

drivers/staging/rdma/hfi1/verbs.h

View file