Re: [Discuss-gnuradio] Question about UHD driver

[Mark McCarron]

I think you are missing the point.  In order to support massive parallelism, data must be duplicated as it comes of the wire and into memory.  Not duplicated in FIFO streams in an application.  The latter is a software implementation of a hardware task and is consuming resources.

It requires hardware and architecture changes to implement properly.

Regards,

Mark McCarron

> Date: Fri, 17 May 2013 18:44:51 +0200
> From: address@hidden
> To: address@hidden
> Subject: Re: [Discuss-gnuradio] Question about UHD driver
>
> > I would tend to agree, but if we do not outline what we require from
> > manufacturers, we will never get it. I would seriously suggest writing
> > a specification and submitting it to Intel, AMD, etc.
> What you want from device manufacturers will never change how networks
> function (by exchanging packeted data with headers and checksums), as
> well as it will never change that latency on a memory bus makes it
> reasonable to always exchange chunks of data; Intel can't do anything
> about that, it's physics...
> An FFT works on a block of samples, so having samples on a
> sample-per-sample-basis won't make it faster.
> However, if you design a DSP program to work on a dedicated chip that
> does *nothing else* than the processing of samples that come directly
> from the ADC, you can of course minimize latency. Sadly, this eliminates
> all reconfigurability, and is not very likely to provide solutions for
> current DSP problems.
>
>