Re: [USRP-users] Dual RX acquisition with E310

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Re: [USRP-users] Dual RX acquisition with E310

Martin Braun via USRP-users
Hi Marcus,

I'm back with the same problem.

In my previous test, UHD versions on desktop and E310 seems to be different. Indeed, UHD version on E310 is 3.9.2-0-unknown (Why "unknown", it's unknown :-) )
I use 3.9.2-0-gf18abe54 on my desktop. So, I cross-compiled the same for E310 and "link it" using sshfs (cf. https://kb.ettus.com/Software_Development_on_the_E310_and_E312)

I was hoping it could be the solution, but ... in fact, no!

The problem is always here.

I launch programs on E310 using "sshfs":
1) "rx_samples_to_file" for individual frontend (i.e., A:A or A:B), it's a success.
2) My "dualRx_samples_to_file" (source code where you didn't find anything wrong), it's failure!
        Datas from 2nd channel (A:B) are good BUT received in buf_ptr[0] instead of buf_ptr[1].
        buf_ptr[1] is filled by 0.

I really try to find a solution by myself but I still need help.

Best regards.

Alex



> Date: Tue, 23 Aug 2016 09:32:59 +0200

> From: Marcus M?ller <[hidden email]>
> To: [hidden email]
> Subject: Re: [USRP-users] Dual RX acquisition with E310
> Message-ID: <[hidden email]>
> Content-Type: text/plain; charset="windows-1252"
>
> Hi Alex,
>
>I'd expect the SD card to be about as fast if not faster than the
>network interface.
>
>I wasn't asking you to test the software on the E310 itself because of
>higher speed, however ? it would have removed the additional complexity
>of using the network mode!
>
>Yes, I've looked through your source code and couldn't find something
>immediately, especially since you said it worked with X310, that
>would've surprised me.
>
>Best regards,
>
>Marcus
>
>
>On 23.08.2016 09:23, alex.dtd--- via USRP-users wrote:
>> Hi Marcus,
>>
>> First of all, thanks for the quick answer.
>>
>> No, I don't test it on the E310 itself because the SD card write speed is too low.
>>
>> I agree that network mode for E310 is just for debug but the rate for the slowest testing dual RX is only 50k/channel <=> 800kB/s ...
>> As I said in my previous message, I checked "/rx_samples_to_file/" @ ~1Msamples/s <=> 8MB/s. So, it's not the problem!
>>
>> The problem should be:
>> 1) I don't use the UHD library well.
>> 2) Pb in UHD library.
>> 3) Pb in E310 firmware.
>>
>> Did you have a look on my source code? Something wrong?
>>
>> Best regards
>>
>> Alex
>>
>>
>> _______________________________________________
>> USRP-users mailing list
>> [hidden email]
>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
_______________________________________________
USRP-users mailing list
[hidden email]
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
Reply | Threaded
Open this post in threaded view
|

Re: [USRP-users] Dual RX acquisition with E310

Martin Braun via USRP-users
On 09/27/2016 08:01 AM, alex.dtd--- via USRP-users wrote:
> Hi Marcus,
>
> I'm back with the same problem.
>
> In my previous test, UHD versions on desktop and E310 seems to be different. Indeed, UHD version on E310 is 3.9.2-0-unknown (Why "unknown", it's unknown :-) )
> I use 3.9.2-0-gf18abe54 on my desktop. So, I cross-compiled the same for E310 and "link it" using sshfs (cf. https://kb.ettus.com/Software_Development_on_the_E310_and_E312 )

Regarding the "unknown" ....

For historical reasons, when uhd builds it insists on inserting the
actual git revision into the version string. But, the OpenEmbedded
recipe for the package on the E310 uses the actual release tarball to
build UHD. So the piece of cmake that runs git to figure out the hash of
the git repo fails because it isn't being run in a git repo.

Philip

>
> I was hoping it could be the solution, but ... in fact, no!
>
> The problem is always here.
>
> I launch programs on E310 using "sshfs":
> 1) "rx_samples_to_file" for individual frontend (i.e., A:A or A:B), it's a success.
> 2) My "dualRx_samples_to_file" (source code where you didn't find anything wrong), it's failure!
> Datas from 2nd channel (A:B) are good BUT received in buf_ptr[0] instead of buf_ptr[1].
> buf_ptr[1] is filled by 0.
>
> I really try to find a solution by myself but I still need help.
>
> Best regards.
>
> Alex
>
>
>
>> Date: Tue, 23 Aug 2016 09:32:59 +0200
>> From: Marcus M?ller <[hidden email]>
>> To: [hidden email]
>> Subject: Re: [USRP-users] Dual RX acquisition with E310
>> Message-ID: <[hidden email]>
>> Content-Type: text/plain; charset="windows-1252"
>>
>> Hi Alex,
>>
>> I'd expect the SD card to be about as fast if not faster than the
>> network interface.
>>
>> I wasn't asking you to test the software on the E310 itself because of
>> higher speed, however ? it would have removed the additional complexity
>> of using the network mode!
>>
>> Yes, I've looked through your source code and couldn't find something
>> immediately, especially since you said it worked with X310, that
>> would've surprised me.
>>
>> Best regards,
>>
>> Marcus
>>
>>
>> On 23.08.2016 09:23, alex.dtd--- via USRP-users wrote:
>>> Hi Marcus,
>>>
>>> First of all, thanks for the quick answer.
>>>
>>> No, I don't test it on the E310 itself because the SD card write speed is too low.
>>>
>>> I agree that network mode for E310 is just for debug but the rate for the slowest testing dual RX is only 50k/channel <=> 800kB/s ...
>>> As I said in my previous message, I checked "/rx_samples_to_file/" @ ~1Msamples/s <=> 8MB/s. So, it's not the problem!
>>>
>>> The problem should be:
>>> 1) I don't use the UHD library well.
>>> 2) Pb in UHD library.
>>> 3) Pb in E310 firmware.
>>>
>>> Did you have a look on my source code? Something wrong?
>>>
>>> Best regards
>>>
>>> Alex
>>>
>>>
>>> _______________________________________________
>>> USRP-users mailing list
>>> [hidden email]
>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com 
>
>
>
> _______________________________________________
> USRP-users mailing list
> [hidden email]
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>


_______________________________________________
USRP-users mailing list
[hidden email]
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
Reply | Threaded
Open this post in threaded view
|

Re: [USRP-users] Dual RX acquisition with E310

Martin Braun via USRP-users
Thanks, Philip

Alex


De: "Philip Balister" <[hidden email]>
À: "alex dtd" <[hidden email]>, [hidden email]
Envoyé: Mardi 27 Septembre 2016 18:27:42
Objet: Re: [USRP-users] Dual RX acquisition with E310

On 09/27/2016 08:01 AM, alex.dtd--- via USRP-users wrote:
> Hi Marcus,
>
> I'm back with the same problem.
>
> In my previous test, UHD versions on desktop and E310 seems to be different. Indeed, UHD version on E310 is 3.9.2-0-unknown (Why "unknown", it's unknown :-) )
> I use 3.9.2-0-gf18abe54 on my desktop. So, I cross-compiled the same for E310 and "link it" using sshfs (cf. https://kb.ettus.com/Software_Development_on_the_E310_and_E312 )

Regarding the "unknown" ....

For historical reasons, when uhd builds it insists on inserting the
actual git revision into the version string. But, the OpenEmbedded
recipe for the package on the E310 uses the actual release tarball to
build UHD. So the piece of cmake that runs git to figure out the hash of
the git repo fails because it isn't being run in a git repo.

Philip

>

> I was hoping it could be the solution, but ... in fact, no!
>
> The problem is always here.
>
> I launch programs on E310 using "sshfs":
> 1) "rx_samples_to_file" for individual frontend (i.e., A:A or A:B), it's a success.
> 2) My "dualRx_samples_to_file" (source code where you didn't find anything wrong), it's failure!
> Datas from 2nd channel (A:B) are good BUT received in buf_ptr[0] instead of buf_ptr[1].
> buf_ptr[1] is filled by 0.
>
> I really try to find a solution by myself but I still need help.
>
> Best regards.
>
> Alex
>
>
>
>> Date: Tue, 23 Aug 2016 09:32:59 +0200
>> From: Marcus M?ller <[hidden email]>
>> To: [hidden email]
>> Subject: Re: [USRP-users] Dual RX acquisition with E310
>> Message-ID: <[hidden email]>
>> Content-Type: text/plain; charset="windows-1252"
>>
>> Hi Alex,
>>
>> I'd expect the SD card to be about as fast if not faster than the
>> network interface.
>>
>> I wasn't asking you to test the software on the E310 itself because of
>> higher speed, however ? it would have removed the additional complexity
>> of using the network mode!
>>
>> Yes, I've looked through your source code and couldn't find something
>> immediately, especially since you said it worked with X310, that
>> would've surprised me.
>>
>> Best regards,
>>
>> Marcus
>>
>>
>> On 23.08.2016 09:23, alex.dtd--- via USRP-users wrote:
>>> Hi Marcus,
>>>
>>> First of all, thanks for the quick answer.
>>>
>>> No, I don't test it on the E310 itself because the SD card write speed is too low.
>>>
>>> I agree that network mode for E310 is just for debug but the rate for the slowest testing dual RX is only 50k/channel <=> 800kB/s ...
>>> As I said in my previous message, I checked "/rx_samples_to_file/" @ ~1Msamples/s <=> 8MB/s. So, it's not the problem!
>>>
>>> The problem should be:
>>> 1) I don't use the UHD library well.
>>> 2) Pb in UHD library.
>>> 3) Pb in E310 firmware.
>>>
>>> Did you have a look on my source code? Something wrong?
>>>
>>> Best regards
>>>
>>> Alex
>>>
>>>
>>> _______________________________________________
>>> USRP-users mailing list
>>> [hidden email]
>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>
>
>
> _______________________________________________
> USRP-users mailing list
> [hidden email]
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>



_______________________________________________
USRP-users mailing list
[hidden email]
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com