On 2021-11-25 12:24, ChunChih Lin wrote:
Hi Marcus,
Thanks a lot for your reply.
So, is there any way that I can check?
Quite apart from the PHY negotiated rate, there's the "benchmark_rate" tool:
UHD Benchmark Rate Allowed options:
--help help message
--args arg single uhd device address args
--duration arg (=10) duration for the test in seconds
--rx_subdev arg specify the device subdev for RX
--tx_subdev arg specify the device subdev for TX
--rx_rate arg specify to perform a RX rate
test (sps)
--tx_rate arg specify to perform a TX rate
test (sps)
--rx_otw arg (=sc16) specify the over-the-wire
sample mode
for RX
--tx_otw arg (=sc16) specify the over-the-wire
sample mode
for TX
--rx_cpu arg (=fc32) specify the host/cpu sample
mode for RX
--tx_cpu arg (=fc32) specify the host/cpu sample
mode for TX
--ref arg clock reference (internal,
external,
mimo, gpsdo)
--pps arg PPS source (internal,
external, mimo,
gpsdo)
--mode arg DEPRECATED - use "ref" and "pps"
instead (none, mimo)
--random Run with random values of
samples in
send() and recv() to
stress-test the
I/O.
--channels arg (=0) which channel(s) to use
(specify "0",
"1", "0,1", etc)
--rx_channels arg which RX channel(s) to use
(specify
"0", "1", "0,1", etc)
--tx_channels arg which TX channel(s) to use
(specify
"0", "1", "0,1", etc)
--overrun-threshold arg Number of overruns (O) which will
declare the benchmark a failure.
--underrun-threshold arg Number of underruns (U) which
will
declare the benchmark a failure.
--drop-threshold arg Number of dropped packets (D)
which
will declare the benchmark a
failure.
--seq-threshold arg Number of dropped packets (D)
which
will declare the benchmark a
failure.
--tx_delay arg (=0.25) delay before starting TX in
seconds
--rx_delay arg (=0.050000000000000003)
delay before starting RX in
seconds
--priority arg (=high) thread priority (high, normal)
Specify --rx_rate for a receive-only test.
Specify --tx_rate for a transmit-only test.
Specify both options for a full-duplex test.
Many thanks
Chun-Chih Lin
On Thu, Nov 25, 2021 at 11:02 AM Marcus D. Leech
<[email protected]> wrote:
On 2021-11-25 10:54, ChunChih Lin wrote:
Hi all,
I have a 10Gb PCIe NIC, Intel X520-DA1, installed on my
desktop and connected to sfp+1 port with SFP+ DAC cable.
I'm trying to verify the speed of the connection.
I launched a iperf3 server on USRP N310, and a client from my
desktop.
The bandwidth is fat from 10Gb. The screenshot below is from my
desktop.
image.png
Bath mtu of interfaces on N310 and my desktop are set to be 9000.
Many thanks
Chun-Chih Lin
_______________________________________________
USRP-users mailing list [email protected]
To unsubscribe send an email [email protected]
You'll not get anywhere close to 10Gig when exchanging packets
with the *CPU* on the N310. You only see the high-bandwidth
rates when streaming samples--the CPU is NOT
involved in that at all.
_______________________________________________
USRP-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
_______________________________________________
USRP-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
