svn commit: r313006 - in head: sys/conf sys/libkern sys/libkern/x86 sys/sys tests/sys/kern
Ben RUBSON
ben.rubson at gmail.com
Wed Mar 22 21:31:43 UTC 2017
> On 01 Mar 2017, at 21:57, Conrad Meyer <cem at freebsd.org> wrote:
>
> Hi Bruce,
>
> On my laptop (Intel(R) Core(TM) i5-3320M CPU — Ivy Bridge) I still see
> a little worse performance with this patch.
Hi Bruce & Conrad,
I gave both patches a try.
It's a real use case, iSCSI throughput.
Both target and initiator running FreeBSD 11, both patched.
I read 100 times 10G from the same raw SSD device.
Results in MB/s.
(top10 is the average of the 10 higher values)
# Without CRC32 :
average : 300.3
std dev : 3.9
top10 : 308.6
max : 311.4
# With software CRC32 :
average : 197.6
std dev : 6.5
top10 : 204.2
max : 204.9
# With Conrad CRC32 :
average : 282.4
std dev : 9.1
top10 : 292.9
max : 293.3
# With Bruce CRC32 :
average : 278.4
std dev : 7.8
top10 : 290.8
max : 292.5
The difference between the 2 patches is let's say almost non significant.
CPU on both sides: Intel(R) Xeon(R) CPU E5-2620 v3 @ 2.40GHz
Ben
More information about the svn-src-all
mailing list