On 16/08/2023 17:26, Johnny Samuels via USRP-users wrote:
I’m trying to remove myself from this mailing list. Sending
unsubscribe to the given email does not remove me from the list. Any
suggestions?
Sent from my iPhone
You sent an "unsubscribe" to:
usrp-users-le...@lists.ettus.com
And it didn't work?
On Aug 16, 2023, at 5:17 PM, Wade Fife <wade.f...@ettus.com> wrote:
You are correct about INGRESS_BUFF_SIZE. It's only used to buffer
data that the stream endpoint receives from another endpoint (e.g.,
data sent from the host computer to a stream endpoint). There's no
extra buffering in the sending stream endpoint. For normal RX where
we stream to a host computer, the computer's memory serves as the buffer.
In your case, if you need extra buffering then I would expect that to
be added to your block. You can change the FIFO sizes in the NoC
shell to add buffering to your block, depending on the type of NoC
shell you're using.
Changing the MTU on the FPGA isn't well tested, so I don't recommend
changing it.
Is the overflow occurring in the radio? If that's the case, then you
probably need additional buffering on the input of your block where
data's received by the radio.
Wade
On Wed, Aug 16, 2023 at 10:45 AM <jmalo...@umass.edu> wrote:
For my application, I am not collecting samples continuously. The
radio block is commanded to stream continuously, but I have a
custom block downstream which “gates” samples in bursts that pass
through. I am able to at least stream data without any overflows
as long as the number of samples the custom block allows is not
too big, which is why I am assuming its a buffer that is too
small. My assumption is as long as my buffers are large enough
and the total number of samples of each burst is less than 10
Gb/s, samples should be able to unload onto the QSFP before the
next burst of samples. This is why I increased the endpoint
buffer sizes.
However, looking over the verilog, it seems INGRESS_BUFF_SIZE
only controls the buffer size of the input going into the
block(before gating), and not the output(after gating), which is
probably why I am still seeing overflows at the same exact rate
even after increasing the buffer size. Is this interpretation
correct, and if so, is it “safe” to control MTU, or will it cause
other problems downstream?
_______________________________________________
USRP-users mailing list -- usrp-users@lists.ettus.com
To unsubscribe send an email to usrp-users-le...@lists.ettus.com
_______________________________________________
USRP-users mailing list -- usrp-users@lists.ettus.com
To unsubscribe send an email to usrp-users-le...@lists.ettus.com
_______________________________________________
USRP-users mailing list --usrp-users@lists.ettus.com
To unsubscribe send an email tousrp-users-le...@lists.ettus.com
_______________________________________________
USRP-users mailing list -- usrp-users@lists.ettus.com
To unsubscribe send an email to usrp-users-le...@lists.ettus.com
