25 MS/s at 16 bits I/Q vs. gr.sizeof_gr_complex

Hello,

I was wondering about a little conflict concerning the sizes of the data
types in GNU Radio.

The UHD driver sends 16 bits I, 16 bits Q samples over the ethernet
interface at 25 MS/s and the return value of “gr.sizeof_gr_complex” is 8
bytes (64 bits).
How does this actually work?
I have looked for some documentation of UHD, but I wasn’t were lucky.

Best regards,
Daniel

On 05/15/2011 09:04 AM, Daniel Bartel wrote:

UHD converts the incoming samples to complex float, scaled into
{-1.0,1.0}.


Principal Investigator
Shirleys Bay Radio Astronomy Consortium

UHD converts the incoming samples to complex float, scaled into {-1.0,1.0}.

Thanks for your answer, I wasn’t sure if it’s “only” a raw convertion.

On 05/15/2011 06:04 AM, Daniel Bartel wrote:
I have looked for some documentation of UHD, but I wasn’t were lucky.

UHD documentation is here:

http://www.ettus.com/uhd_docs/manual/html/

Matt

Daniel,

What hardware setup do you use to achieve 25MS/s? I wasn’t able to
get stable connection with more then 16MS/s, and even at 16MS/s I had
drops.

On Sun, May 15, 2011 at 17:04, Daniel Bartel
[email protected] wrote:


Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page


Regards,
Alexander C…

On 16/05/2011 10:24 AM, Alexander C. wrote:

Daniel,

What hardware setup do you use to achieve 25MS/s? I wasn’t able to
get stable connection with more then 16MS/s, and even at 16MS/s I had
drops.
I use a 6-core AMD Phenom II X6 1090T with 8GB of 1333MHz memory for my
radio-astronomy work, and it can process (just barely) 25Msps.

Hello Alexander,

in fact my question was only a theoretical one, because I read in the
mailinglist about the ethernet connection and that 2 x 16Bit x 25MS/s =
800Mbit/s are used.
Concerning my hardware setup I’ve to look tomorrow, because it’s not
actually my pc.

Am 15.05.2011 um 16:07 schrieb Marcus D. Leech:

UHD converts the incoming samples to complex float, scaled into {-1.0,1.0}.
Does this mean, that there is a loss in accuracy, because a 32 bit data
type (one pair of gr_complex) is converted to a 16 bit data type
(float)?

Best regards,
Daniel

Am 16.05.2011 um 16:24 schrieb Alexander C.:

On Mon, May 16, 2011 at 19:04, Marcus D. Leech [email protected]
wrote:

On 16/05/2011 10:24 AM, Alexander C. wrote:

Daniel,

What hardware setup do you use to achieve 25MS/s? I wasn’t able to
get stable connection with more then 16MS/s, and even at 16MS/s I had
drops.

I use a 6-core AMD Phenom II X6 1090T with 8GB of 1333MHz memory for my
radio-astronomy work, and it can process (just barely) 25Msps.

I see. That’s far better then a laptop I used for my experiments.
Thank you for the information.


Regards,
Alexander C…