[PATCH libdrm 0/2] improve reuse implementation

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



From: "Xiong, James" <james.xiong@xxxxxxxxx>

This series 1) align the reuse buffer size to page size instead. The
goal is to reduce memory penalty (up to 25% when reuse is enabled )
while maintain similar performance. A potential overhead is: since a
bucket now contains cached buffers with difference sizes, we need go
through the list to search for a suitable buffer for reuse. However
since the buckets get cleaned up regularly and the older cached buffer
are removed, the list should not be long and the performance impact
should be minor.

2) fix a bug that a bucket size instead of the requested size was ALWAYS
used for buffer allocation even when bo_reused is disabled.

Xiong, James (2):
  intel: align reuse buffer's size on page size instead
  intel: allocate buffer with the requested size when reuse is disabled

 intel/intel_bufmgr_gem.c | 185 +++++++++++++++++++++++++----------------------
 libdrm_lists.h           |   6 ++
 2 files changed, 105 insertions(+), 86 deletions(-)

-- 
2.7.4

_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel




[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux