On 04/11/2024 20:12, dhpanch...@gmail.com wrote:
I got it work. It seems RT_RUNTIME_SHARE disabled was the culprit. I
re-enabled it using these instructions and the benchmark worked
without packet drops or underruns:
*Underruns Every Second with DPDK + Ubuntu*
With Linux kernels 5.10 and be
I got it work. It seems RT_RUNTIME_SHARE disabled was the culprit. I re-enabled
it using these instructions and the benchmark worked without packet drops or
underruns:\
\
**Underruns Every Second with DPDK + Ubuntu**
With Linux kernels 5.10 and beyond, we have observed periodic underruns on
sys
This is because a peek command will always wait unti it's finished, whereas
the other commands ("poke" commands) will stack in the command FIFO. If you
add the blue lines (the set_command_time() is redundant BTW) then the call
will wait until all the previous GPIO commands are finished, then it wil
On Thu, Oct 31, 2024 at 3:25 PM wrote:
> I didn’t quite resolve this, but I think I narrowed it down to some bug
> when switching between DPDK and non-DPDK drivers... I also switched back to
> the default HG image for troubleshooting. After I bind the vfio-pci driver,
> rebinding the i40e driver
Just to update on my previous question. I also tried to update
"rfnoc_rx_streamer.cpp" by changing function *spp*4 * in the*
"add_property_resolver function"* as follows:
*add_property_resolver({atomic_item_size_in, mtu_in, type_in},
{},[&ais = *atomic_item_size_