Hello,
We are using one of our T2080RDB (running SDK-1.9 / linux-3.12.37) to receive a continous stream of 9k jumbo packets (UDP) on the 10G link (fman1-mac9).
Compared to SDK-1.6 where packets were lost, SD-1.9 behaves a lot better.
However we are experiencing the following problems:
Is there any way how we can receive the packets in order?
Are there tuning possibilities for our kind of workloads to get the software interrupt rate down?
Thank you in advance, Clemens
Some further information: For 63,5TB data (consisting of 9k jumbo frames), 4,35G interrupts were generated - which translates to 1 interrupt per ~15kb of data.
Is this to be expected?
root@t2080rdb:~# cat /proc/interrupts
CPU0 CPU1 CPU2 CPU3 CPU4 CPU5 CPU6 CPU7
124: 0 0 0 0 0 0 0 546099313 OpenPIC 124 Level QMan portal 7
125: 0 0 0 0 0 0 0 0 OpenPIC 125 Level BMan portal 7
126: 0 0 0 0 0 0 555860282 0 OpenPIC 126 Level QMan portal 6
127: 0 0 0 0 0 0 0 0 OpenPIC 127 Level BMan portal 6
128: 0 0 0 0 0 545426042 0 0 OpenPIC 128 Level QMan portal 5
129: 0 0 0 0 0 0 0 0 OpenPIC 129 Level BMan portal 5
130: 0 0 0 0 554352922 0 0 0 OpenPIC 130 Level QMan portal 4
131: 0 0 0 0 0 0 0 0 OpenPIC 131 Level BMan portal 4
132: 0 0 0 545187734 0 0 0 0 OpenPIC 132 Level QMan portal 3
133: 0 0 0 0 0 0 0 0 OpenPIC 133 Level BMan portal 3
134: 0 0 554957818 0 0 0 0 0 OpenPIC 134 Level QMan portal 2
135: 0 0 0 0 0 0 0 0 OpenPIC 135 Level BMan portal 2
136: 0 546313678 0 0 0 0 0 0 OpenPIC 136 Level QMan portal 1
137: 0 0 0 0 0 0 0 0 OpenPIC 137 Level BMan portal 1
138: 556060144 0 0 0 0 0 0 0 OpenPIC 138 Level QMan portal 0
139: 0 0 0 0 0 0 0 0 OpenPIC 139 Level BMan portal 0
*bump*