Hi Glen,
As far as I can remember, the bus address of the 10GbE blocks are
dynamically allocated. In either case, you should be able to figure
it out by going into XPS_iBOB_base/ppc405_1/include/xparameters.h,
which is the end-all, be-all index of address for anything connected
to the PowerPC.
If Port0 is mapped to x40000000, though, I think Port1 would be
mapped at x40004000. Wherever it ends up, I'm betting it'll be fairly
consistent, as not that many things wind up on the PLB bus on the
IBOB other than 10GbE.
Thanks,
Henry
John Ford wrote:
Hi Henry and Terry,
Thanks for the info on configuring the 10 GBe.
One question concerning configuring
the IBOB 10 GBe memory addresses:
For use of XAUI 0 we have to set some values in
memory locations. Our question is, essentially,
where is the document describing what memory locations
that need to be set for XAUI 1?
Hi Glen.
I don't know if it's written anywhere, but you just change the base
address of the controller. you can use the tinyshell to find the base
address. The offsets are the same for both ports
Here's something I chopped out of one of your parspec scripts:
regwrite reg_ip 0xc0a8030a
regwrite reg_10GbE_destport0 4001
---> setb x40000000 <-------- Change this.
writeb l 0 x00000060
writeb l 4 xdd47e301
writeb l 8 x00000000
writeb l 12 xc0a8030b
writeb b x16 x0f
writeb b x17 xa1
When Dong was here, I think we had his design working on both ports if you
can find his scripts.
John
We think the jumpers are set correctly.
Thanks
Glen
Hi Glen,
The jumper problem is actually a bit of a long shot, but worth
double-checking.
The background is that the IBOB (and BEE2) is designed to support
both 10GbE (10Gb/s) and InfiniBand (8Gb/s). The differences in
data rates are enabled by using different reference clocks for
the transceivers, 156.25MHz for 10Gb/s, and 125.00MHz for 8Gb/s.
The IBOB, due to a lack of clock input buffers, requires manual
selection of these clock inputs using on-board jumpers at J11/12
and J13/14.
Steps 3.2 and 11.5 in the IBOB test procedure
(http://casper.berkeley.edu/memos/ibob_test_proc.pdf)
is a good reference for these jumpers. You want to make sure it's
set to use 156.25MHz as in step 11.5
I think the only optical transceiver we've had experience with
on the BEE2 and IBOB is the Fujitsu CX4 converter
(http://www.fujitsu.com/downloads/MICRO/fcai/connectors/fpd-010r008-0e.pdf)
That one's been used pretty successfully by a cognitive radio
research group at BWRC.
Glen Langston wrote:
Hi Andrew and Henry,
Did you happen to run across any documentation
on the shared memory locations for the IBOB using
the 2nd XAUI port as a 10 Gbe interface?
Also we are preparing to purchase some fiber
modems that will allow us to transmit the
10 Gbe over distances of order 600ft.
Which fiber transmitters/receivers do you
have experience with?
Thanks
Glen
henry,
can you please answer this with a cc to casper-lists,
so we can capture this query and answer for others.
(casper emails are now captured and go on the wiki
automatically, thanks to billy).
thanks,
dan
Glen Langston wrote:
Hi Henry,
Thanks for your help with the parspec design yesterday.
The ADC 1, XAUI 1 design built, with your changes.
However, on further consideration, we may need to make some
changes in how we configure the 10 Gbe for use of XAUI 1.
It appears the configuration is memory mapped. We have
a text file that we feed IBOB with LWIP commands to
setup the various communications parameters.
(One is attached).
So, at some point someone mentioned that we may have
to jumper the XAUI connectors to get the 10 Gbe to work.
I've looked at the documentation, but do not see the
location of these jumpers. Also, we're guessing that
there are other memory locations for the XAUI 1 device.
Is there a document describing these locations?
Thanks
Glen
To send the commands to the IBOB, at the linux prompt we run
the following command sequence:
python ibob5.py
This sends the commands in file
parspecTofu.txt
but ignores comments, so we can document the configuration file.
We think this is a good and convenient scheme for use with the ibobs.
