[hpsdr] Hermes_Protocol_2_v10.4 initial Gbps firmware available to try - beta test version

[Isaac Weksler]

Hi, Joe,

Thank you for the prompt answer and detailed response. I certainly did not
mean to address my question to you personally and by no means I expect you
or any other developer to have any obligations towards anybody using the
great achievement called HPSDR.
I addressed the question to you since, from your announcement, I assumed
that you are in charge of the Protocol 2 development in general and would
know about the plans in this direction. I am not very knowledgeable who does
what in the HPSDR project. However, I am extremely grateful and appreciative
of the efforts the volunteer developers put into this project and I think I
speak here for all the hams who have chosen to participate in HPSDR project,
would that be as contributors or by operating the HPSDR products for our
great hobby.
Now, regarding your explanation, I understand exactly what you mean. I just
recall discussions that took place some time ago about the possibility of
upgrading the Atlas bus protocol by utilizing the unused lines for
increasing its throughput. Generally, there is some advantage in using
separate boards for different functions as opposed to the architecture where
everything is crammed into one board. Personally, I decided to go for
separate boards on Atlas bus instead of a single board transceiver like
Hermes because it allowed greater flexibility in experimenting. Besides,
higher performance may sometimes be achieved if the development effort is
focused on each of the boards separately.
Of course you are right about FPGAs and some of the other components used in
Mercury, Pennylane and Metis – they are not the latest. But that's the fate
of all boards and devices – electronics advances extremely fast. Does that
mean that we should throw away all the stuff we have and get new one every
time a new component appears at the market?
Of course I do not mean you personally. Just expressing my opinion.

73 Isaac 4Z1AO

 

-----Original Message-----
From: Joe Martin [mailto:k5so at k5so.com] 
Sent: Tuesday, May 2, 2017 5:53 PM
To: Isaac Weksler <mailto:iweksler at bezeqint.net>
Cc: hpsdr <mailto:hpsdr at lists.openhpsdr.org>
Subject: Re: [hpsdr] Hermes_Protocol_2_v10.4 initial Gbps firmware available
to try - beta test version

Hi Isaac, 

As I have posted previously on this reflector,  I personally do not have
specific plans for developing additional code for Atlas-based systems, for
two principal reasons: 

1) In my personal view, the Atlas-based systems no longer fall into the
category of “high performance” because we already have reached the maximum
data rate limits for the Altlas bus design and those limits are
significantly lower in performance than any of the single board transceiver
alternative options we now have available.  Indeed, the last change I
implemented for the Atlas-based systems was to enable it to run at 384ksps
and to do so it was necessary to pass the I and Q values from the Mercury
board over two separate bus lines (one line for I and one line for Q from a
single Mercury board) to Metis in order to remain below the maximum data
rate limitation for the bus lines.  While this approach works it is hardly
an appropriate scheme to use generally I think and by doing so we have used
all the bus lines so we have effectively reached the performance limit (in
terms of useable data rates) for that platform as it currently exists, in my
opinion.  

2) The FPGAs used in the Atlas-based systems are elderly Cyclone II
(Penny/PennyLane) and Cyclone III (Mercury/Metis), neither of which are
supported any longer in current Quartus versions, so to work on them it is
necessary to use early verisions of Quartus which do not have the advances
that have been employed in current Quartus versions.  

While the two situations mentioned above do not preclude developing 100BaseT
Protocol 2 code for the Atlas-based system it is not something that I
personally am very interested in spending time doing.  Other developers may
wish to do it though; that’s just fine and I hope someone does it for you,
but your question was directed to me personally; this is my personal
response.  The answer to your question to me is then, no, I have no specific
plans to do that development work personally but perhaps someone else in the
group does have such plans of which I am unaware.

I recall that Phil mentioned at one time, before his day job and personal
committments recently increased to further limit his time available for
HPSDR development issues, that he intended to examine the possibilities of
developing Protocol 2 code for Atlas-based systems, even though it would
necessarily need to be restricted to 100T data rates of course, but it is
conceivable that he may have changed his mind about that, I don’t know. 
Neither he nor anyone else has mentioned anything further along those lines
to me personally nor publicly that I know.  

The statements above reflect my personal view and do not reflect any
official position for the HPSDR group, of course.  If you intended the query
to be a general question to all developers then you should direct the
question accordingly, I think, not simply directed to me.  

73, Joe K5SO