Re: [Discuss-gnuradio] GNURadio OOT vector input/output with different size

[Marcus Müller]

Ali,

Is that really your forecast? That won't work. You'll need to implement it.
Unless you actually meant to do consume_each(noutput_items) AND return(noutput_items), because then you shouldn't be using a general gr::block, but gr::sync_block, which saves you the hassle of implementing a forecast and of explicitly consuming items.

Best regards,
Marcus

On 07/21/2017 08:20 AM, Ali wrote:

#ifdef HAVE_CONFIG_H

#include "config.h"

#endif

#include <gnuradio/io_signature.h>

#include "block_name_impl.h"

#include "function.h"

namespace gr {

 namespace module {

  

   block_name::sptr

   block_name::make(float parameter)

   {

     return gnuradio::get_initial_sptr

(new block_name_impl(parameter));

   }

   /*

    * The private constructor

    */

int myints[] = {4000,4000,4000,4000,4000,4000,4000,4000};

const std::vector<int> inputs (myints, myints+sizeof(myints)/sizeof(float));

int myints2[] = {4000,4000,4,4};

const std::vector<int> outputs (myints2, myints2+sizeof(myints2)/sizeof(float));

 block_name_impl::block_name_impl(float parameter)

  : gr::block("block_name",

gr::io_signature::makev(8,8,inputs),

gr::io_signature::makev(4,4,outputs)),

parameter2(parameter)

 {}

   /*

    * Our virtual constructor

    */

   block_name::~block_name_impl()

   {

   }

   void

   block_name_impl::forecast(int noutput_items, gr_vector_int &ninput_items_required)

   {

     /* <+forecast+> e.g. ninput_items_required[0] = noutput_items */

   }

   int

   block_name_impl::general_work (int noutput_items,

                            gr_vector_int &ninput_items,

                            gr_vector_const_void_star &input_items,

                            gr_vector_void_star &output_items)

   {

     const float *in1 = (const float *) input_items[0];

     const float *in2 = (const float *) input_items[1];

     const float *in3 = (const float *) input_items[2];

     const float *in4 = (const float *) input_items[3];

     const float *in5 = (const float *) input_items[4];

     const float *in6 = (const float *) input_items[5];

     const float *in7 = (const float *) input_items[6];

     const float *in8 = (const float *) input_items[7];

     float *out1 = (float *) output_items[0];

     float *out2 = (float *) output_items[1];

     float *out3 = (float *) output_items[2];

     float *out4 = (float *) output_items[3];

     // Do <+signal processing+>

     // Tell runtime system how many input items we consumed on

     // each input stream.

function(in1,in2,in3,in4,in5,in6,in7,in8,parameter2,out1,out2,out3,out4);

consume_each(noutput_items);

return noutput_items;

     } 

  }

}

  

The "function line" may be inside a for, I am not sure. I am not sure of the bold lines also. What is your overall comments?

2017-07-21 8:52 GMT+03:00 Ali <address@hidden>:

Hi,

Patrick's code does not work since I think I am using c++98. I dont know how to enable c++11. Anyway for c++98 I wrote the followings(I found on the internet):

-->

int myints[] = {4000,4000,4000,4000,4000,4000,4000,4000};

const std::vector<int> inputs (myints, myints+sizeof(myints)/sizeof(float));

int myints2[] = {4000,4000,4,4};

const std::vector<int> outputs (myints2, myints2+sizeof(myints2)/sizeof(float));

-->

makev(8,8,inputs) and makev(4,4,outputs)

It is compiled after "sudo make" command.

But when I wrote "gr_modtool makexml function", I got the following error(... means I skipped writing these lines).

...

Making GRC bindings for lib/function_impl.cc...

tbi

Error: Can't parse input signature.

tbi

Error: Can't parse output signature.

...

 if iosig[inout]['max_ports'] == '-1':

Key error: 'in'

I will post my whole code later.

2017-07-21 1:14 GMT+03:00 Patrick Sathyanathan <address@hidden>:

Try this instead:

const std::vector<int> inputs{4000,4000,4000,4000,4000,4000,4000,4000};

and make sure you compile with c++11 enabled.

--Patrick

---

From: Discuss-gnuradio <discuss-gnuradio-bounces+wpats=address@hidden> on behalf of Ali <address@hidden>
Sent: Thursday, July 20, 2017 12:13 AM
To: address@hidden
Subject: Re: [Discuss-gnuradio] GNURadio OOT vector input/output with different size

 

Hi,

I dont have any background about C++. So I could not generate the std::vector<int> with the desired content and could not fınd anything on the internet. 

const std::vector<int> inputs[] = {4000,4000,4000,4000,4000,4000,4000,4000};

does not work. I am getting the following error: 

"conversion from 'int' to non-scalar type 'std::vector<int>' requested"

Best,

Ali

 

2017-07-18 11:51 GMT+03:00 Marcus Müller <address@hidden>:

Hi Ali,

I think this should also work, shouldn't it?

no, that's why I explained what you need to do (use makev).

Does this part need to be changed since I am working with the vectors?

No, not really. The vectors are still only consecutive numbers in memory.

Best regards,

Marcus

On 18.07.2017 07:51, Ali wrote:

Hi,

1- Actually I am using the followings

make(8,8,1000*sizeof(float))

make3(4,4,1000*sizeof(float),1000*sizeof(float),4*sizeof(float))

I think this should also work, shouldn't it?

2- Under the general work function I am using the followings:

const float *in1 = (const float *) input_items[0]

...

float *out1 = (float *) output_items[0]

Does this part need to be changed since I am working with the vectors?

Thanks,

Ali

2017-07-17 16:42 GMT+03:00 Marcus Müller <address@hidden>:

Hi Ali,

So, you want one block with:

• 8 inputs, itemsize0…7 = 1000*4B = 4000B
• 4 outputs, itemsize0=itemsize1=4000B, itemsize2=itemsize3=16B

Correct?

You need to generate two std::vector<int> with content

{4000,4000,4000,4000,4000,4000,4000,4000}

and

{4000,4000,16,16}

respectively, and use gr::io_signature::makev(int min_streams, int max_streams, vector) to generate the io_signatures[1] that you use in your block's constructor. The question whether your block should be a general, or a sync block, is independent from the item sizes of the in and outputs, but depends on whether there's always a fixed ratio of produced output items to consumed input items, as explained in [2].

Best regards,

Marcus

[1] https://gnuradio.org/doc/doxygen/classgr_1_1io__signature.html#a99e0f9e8de8e7ce16ed92d9f2655e66c
[2] https://wiki.gnuradio.org/index.php/Guided_Tutorial_GNU_Radio_in_C%2B%2B#4.3.2_Specific_block_categories

On 07/17/2017 03:31 PM, Ali wrote:

Hi to all,

I want to design my own OOT module with the following I/O:

8 inputs (length of 1000 and each element is type of float)

2 outputs (length of 1000 and each element is type of float)

2 outputs (length of 4 and each element is type of float)

I used general type block but I could not get the desired outputs. Do you suggest other type of blocks? Is there any example similar to this work that I can study on? Can you suggest a module name or a link?

Best,

Ali

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

_______________________________________________ Discuss-gnuradio mailing list address@hidden https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

_______________________________________________ Discuss-gnuradio mailing list address@hidden https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio