Hello Chintan:

Yes, on the B210, in 1x1 mode, the maximum sampling rate is 61.44 Msps, and
in 2x2 mode, the maximum sampling rate is 30.72 Msps.

The tweak/modification that you mention has not been implemented for the
B210.

--Neel Pandeya




On 9 September 2018 at 13:20, Chintan Patel via USRP-users <
[email protected]> wrote:

> Hi Marcus,
>
> Thanks for the clarification.  So the max rate for two channels on the
> B210 is 30.72Msps, right? Reading the 9361 manual, it seems that if one
> wants to operate in RX-only (dual-port half-duplex) mode, the 61.44Msps
> rate might be achievable (at the expense of no TX). Any idea if this
> modification has been tried out on the B210? It would be an invasive change
> since the AD9361 digital interface in the FPGA would need tweaks, but any
> thoughts on if it was theoretically possible or not.
>
> Chintan
>
>
> On Sun, Sep 9, 2018 at 11:33 AM Marcus D. Leech via USRP-users <
> [email protected]> wrote:
>
>> On 09/09/2018 11:21 AM, Chintan Patel via USRP-users wrote:
>> > Hi,
>> >
>> > Two questions on the B210.
>> >
>> > 1. Generally speaking, are there are known issues transporting two RX
>> > channels at 61.44Msps over USB 3.0 on the B210. I know from a
>> > theoretical standpoint, the 5Gbps capacity of USB 3.0 has enough
>> > capacity to transport the dual-channel 16-bits i/q @ 61.44Msps. I also
>> > know that whether the host can keep up with this throughput depends on
>> > host configuration, load-balancing etc., but wanted to get a general
>> > sense of which bucket it falls under: a) i
>> > quite-typical-and-routine-scenario, b)
>> > possible-but-in-a-particular-context, or c) pushing-the-capabilities
>> >
>> > 2. Taking this thought a bit further - (our application needs 4
>> > channels), if we had a powerful dedicated host machine, how
>> > feasible/unrealistic is to to have a configuration where the host is
>> > connected to two B210, via separate USB 3.0s and ingesting 4 channels
>> > worth of RX data at 61.44Msps. From a UHD driver standpoint, I am
>> > assuming this would just entail interfacing with two separate
>> > instances of streamers, but is there any pit-fall in the single
>> > host-to-two-B210 approach?
>> >
>> > Thanks,
>> > Chintan
>> >
>> It's not a driver or host-beefiness issue with B210.  It's the way the
>> data clocks work on the AD9361 chip, and the interface to the FPGA.
>>
>> There's simply no way to make the AD9361 chip move two channels of data
>> out of itself (or, into itself) at the maximum data rate.
>>
>>
>>
>> _______________________________________________
>> USRP-users mailing list
>> [email protected]
>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>>
>
> _______________________________________________
> USRP-users mailing list
> [email protected]
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>
>
_______________________________________________
USRP-users mailing list
[email protected]
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Reply via email to