discuss-gnuradio
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] nVidia's Tegra 4 has SDR - the i500 LTE soft mode


From: Albert Chun-Chieh Huang
Subject: Re: [Discuss-gnuradio] nVidia's Tegra 4 has SDR - the i500 LTE soft modem from Icera
Date: Mon, 14 Jan 2013 22:39:05 +0800
User-agent: Gnus/5.110018 (No Gnus v0.18) Emacs/24.2 (darwin)

Hi, Lin,

According to Icera's previous product lines, there is no any
documentation for instruction sets. I think their market is the same as
Qualcomm's, i.e. cell phone manufactorurs. TI has a digital signal
processor C6670, which targets base stations. It contains some
coprocessors, e.g. turbo encoder/decoder, FFT coprocessors, to perform
signal processing tasks. TI provides detailed documentation for DSP
instruction set as well as these coprocessor configuration. IMHO, TI
C6670 is more suitable for GNU Radio guys to DIY something. C6670 EVM
has Gigabit Ethernet interface to connect to USRP N2xx and is sold at
the price of USD$599. You can find information for C6670 at
http://processors.wiki.ti.com/index.php/C6670

IMHO, to do SDR on "generalized" CPU like x86 is very difficult to
achieve low power consumption if the targeted standard is cellular
communication. I would guess that Icera's approach is using some SIMD
processors with instruction set specialized for signal processing tasks,
such as FFT, Turbo codec, Rake receivers. Optimizing programs with that
kind of instruction set requires long-time training and careful
tuning. And I guess they use assembly to write these programs in order
to save memory footprint and hence reduce die size of the chip. But
that's just my guess.

Software radio, as I imagine and expect, would be very easy to program
in high level language with a lot of flexibility and many
already-existed components. If computing power is not enough to perform
real-time communication on a single computer, it is reasonable to split
tasks among several computers. On ICE's website, it is compared to
CORBA, which is a distributed computing framework/service. By
introducing ICE into GNU Radio, as I expect happily, would make
distributed SDR possible, that means if we need more computing power for
real-time communication, we might be able to add more computers with
careful splitting tasks among these computers! And these programs are
written in high level languages! That makes developing communication
more enjoyable than writing and tuning assembly code!

Cheers,

Albert

Lin HUANG <address@hidden> writes:

> Hi Albert,
>
> If that is as you said, Icera won't open the instruction set and develop
> tool, and all the software are encrypted. Then this chipset is not suitable
> for people like GNU Radio guys to DIY something. So, what is the major
> market of this chipset? Cellphone manufactor? Let them to develop more
> diverse products?
>
> I paid attention to Icera's solution for a long time. I hope there will be
> a small chipset which can be used as CPU plus USRP, with low power
> comsumption, suitable for mobile terminal. Based on your knowledge on
> industry, do you think  when and what kind of solution may come to market?
> What is 'a scalable computer farm
> that can do distributed SDR' that you said?
>
> Regards
> Lin
>
> 2013/1/9 Albert Chun-Chieh Huang <address@hidden>
>
>> I'd rather spend time to build a scalable computer farm
>> that can do distributed SDR
>>
> Hi Albert,
>
> If that is as you said, Icera won't open the instruction set and
> develop tool, and all the software are encrypted. Then this chipset is
> not suitable for people like GNU Radio guys to DIY something. So, what
> is the major market of this chipset? Cellphone manufactor? Let them to
> develop more diverse products?
>
> I paid attention to Icera's solution for a long time. I hope there
> will be a small chipset which can be used as CPU plus USRP, with low
> power comsumption, suitable for mobile terminal. Based on your
> knowledge on industry, do you think  when and what kind of solution
> may come to market? What is 'a scalable computer farm
> that can do distributed SDR' that you said?
>
> Regards
> Lin
>
> 2013/1/9 Albert Chun-Chieh Huang <address@hidden>
>
>     I'd rather spend time to build a scalable computer farm
>     that can do distributed SDR
>
>
>

-- 
Albert Chun-Chieh Huang(黃俊傑)
Blog: Random Notes, http://alberthuang314.blogspot.com/



reply via email to

[Prev in Thread] Current Thread [Next in Thread]